60/349,339 filed January 22,2002 ; United States patent application serial no. 60/349,371 filed January 22,2002 ; and United States patent application serial no. 60/393,385 filed July 5,2002.

FIELD OF THE INVENTION The present invention relates to sequence analysis, and more particularly, to apparatus, methods, computer-readable media, computer programs, signals and data structures for protein sequence analysis.

BACKGROUND OF THE INVENTION Methods are known for comparing the amino acid sequences of proteins to identify indel (sertion/deletion) associated sequences. Such methods are described for use in identifying peptides, peptide analogs, and peptidomimetics that have a high probability of modulating or mimicking the activity of a target protein. For example, in PCT international patent application publication no. WO 97/01578, methods are described for identifying molecules that affect biological activity of a target protein in which information regarding the location of an indel in a target protein is used to identify nearby fragments.

Protozoan pathogens are responsible for a wide variety of disease, typically tropical in nature. These pathogens exist as intracellular parasites in a host and include organisms that appear to be protozoan in nature such as P7zeumocystis. An example of

such protozoan pathogens is the genus Leishmania, which causes a spectrum of tropical and subtropical diseases known as the leishmaniases. These consist of several forms including cutaneous, mucocutaneous, visceral and diffuse cutaneous leishmaniasis. The incidence of such tropical diseases has been on the rise due to multiple factors including the AIDS epidemic, increased international travel, lack of effective vaccines, difficulty in controlling vectors, and the development of resistance to chemotherapy.

Leishmania live as either extracellular, flagellated promastigotes in the digestive tract of their sand fly vector or as non-flagellated amastigotes within macrophages, where they survive and replicate within phagolysosomes. During both the innate and acquired immune responses, macrophages respond to extracellular signals to become activated for enhanced antimicrobial activity. This is a critical requirement leading to the elimination of intracellular pathogens. However, recent findings have shown that leishmania and other intracellular pathogens have developed mechanisms to interfere with cell signaling pathways thereby preventing macrophages from becoming effectively activated [Reiner NE. (1994) Immunol. Today 15: 374-381 ; Nandan, D. et al. (2000) Leukocyte Biology 67: 464-470]. As a result, these organisms are able to survive and successfully multiply within the otherwise hostile intracellular milieu of macrophages.

L. donovani is the major causative agent of human visceral leishmaniasis. This disease is progressive and often fatal if untreated. Recently, it has been shown that macrophages infected with L. donovani show a phenotype of impaired cell signaling and cell deactivation. For example, interferon-y signaling through the Jak-Statl pathway [Nandan D. and Reiner NE. (1995) Infect. Immun. 63: 4495-4500] and mitogen-activated protein kinase signaling leading to iNOS induction and c-FOS expression were observed to be markedly attenuated in leishmania infected cells [Nandan D. et al. (1999) Infection and Immunity 67 : 4055-4063]. Of significant interest was the finding that the deactivated phenotype was reversed in cells that had been incubated with the protein tyrosine phosphatase (PTP) inhibitor sodium orthovandate prior to infection [Nandan D. et al.

(1999) Infection arad Inamufaity 67 : 4055-4063]. Several lines of evidence have recently converged to demonstrate a role for the Src-homology 2 (SH2) domain containing protein tyrosine phosphatase-1 (SHP-1) in the pathogenesis of infections with leishmania [Nandan D. et al. (1999) Infection and Immunity 67: 4055-4063; Olivier M. et al. (1998) J.

Biol. Chem. 273: 13944-13949; Blanchette J. et al. (1999) Eu7 : J. Immunol. 29: 3737-3744;

Forget G. et al. (1999) Jou7nal of Leutocyte Biology Supplement 31 : 1999]. In particular, SHP-1 has been shown to become activated in leishmania infected cells [Nandan D. et al.

(1999) Infection and Immunity 67: 4055-4063 ; 6] and leishma7lia infection is attenuated under conditions of SHP-1 deficiency [Forget G. et al. (1999) Journal of Leukocyte Biology Supplement 31: 1999]. Moreover, it has recently been shown that a conventional anti-leishmanial agent used clinically, sodium stibogluconate, is an inhibitor of SHP-1 [Pathak MK. and Yi T. (2001) JImmunol. 167: 3391-3397].

SUMMARY OF THE INVENTION It was surprising to discover that essential pathogen proteins can contain insertions or deletions compared to their host homologues conferring structural differences that may be exploited as drug targets specific for the pathogen. Proteins that are essential to pathogen survival can contain discrete, heretofore unrecognized indels which differentiate them from their otherwise highly related mammalian homologues. Up to now, because of their high degree of relatedness and because many of the host proteins were essential for survival of both the host and the pathogen, these proteins have not been candidate drug targets.

In accordance with one aspect of the invention, there is provided a method of protein sequence analysis, the method including identifying, from among a first group of aligned protein sequences having at least a threshold degree of alignment with a first protein sequence, a second group of at least one of the aligned sequences having at least one insertion or deletion ("indel") relative to the first protein sequence satisfying a predefined condition.

The first protein sequence may include a protein sequence of a pathogen, and the predefined condition may be indicative of the potential presence of a drug target for the pathogen associated with the at least one indel.

At least one of the first group of aligned protein sequences may include a protein sequence of a host organism associated with the pathogen.

The predefined condition may include a minimum length of the at least one indel.

The minimum length may include a minimum length of at least four contiguous amino acids.

Alternatively, or in addition, the predefined condition may include a maximum length of the at least one indel.

The maximum length may include a maximum length of SO contiguous amino

acids.

The predefined condition may include a predejmed range of length of the at least one indel. The predefined range may include a range of four to 50 contiguous amino acids.

The method may further include comparing a plurality of protein sequences to the first protein sequence to identify the first group of aligned protein sequences having at least me threshold degree of alignment with the first protein sequence.

The threshold degree of alignment may be at least 70% alignment between each of the aligned protein sequences and the first protein sequence. More particularly, the threshold degree may be at least 80% alignment, or may be at least 90 % alignment, for example.

The method may further include successively selecting each one of a plurality of protein sequences as the first protein sequence, and repeating the identifying for the each one of the plurality of protein sequences- The method may further include storing the second group of aligned sequences in a storage medium.

In accordance-with another aspect of the invention, there is provided an apparatus for protein sequence analysis. The apparatus includes at least one processor circuit configured to perform the methods disclosed herein.

In accordance with another aspect of the invention, there is provided a computer program including code means that when executed on at least one processor circuit carry out the methods disclosed herein.

In accordance with another aspect of the invention, there is provided a computer program on a carrier carrying code that when executed on at least one processor circuit directs the at least one processor circuit to carry out me methods disclosed herein.

In accordance with another aspect of the invention, there is provided a computer- readable medium providing code segments for directing at least one processor circuit to carry out the methods disclosed herein.

In accordance with. another aspect of the invention, there is provided a signal embodied in a transmission medium including code segments for directing at least one processor circuit to carry out the methods disclosed herein.

In accordance with another aspect of the invention, there is provided a signal embodied in a carrier wave including code segments for directing at least one processor circuit to carry out the the methods disclosed wherein.

In accordance with another aspect of the invention, there is provided a method including generating a signal including code segments for directing at least one processor circuit to carry out the methods disclosed herein.

In accordance with another aspect of the invention, there is provided a method including propagating a signal including code segments for directing at least one processor circuit to carry out the methods disclosed herein.

In accordance with another aspect of the invention, there is provided an apparatus for proteins sequence analysis, the apparatus including : <BR> <BR> <BR> <BR> a) means for receiving data lepresenting, a first group of aligned protein sequences having at least a threshold degree of alignment with the first protein sequence ; and b) means for identifying, from among the Erst group) a second group of at least one of the aligned sequences having at least one insertion or deletion ("indel") relative to the first protein, sequence satisfying a predefined condition.

In accordance with another aspect of the invention, there is provided a method of protein sequence analysis, the method including : a) receiving identifications of aligned protein sequences at least some of which have at least a threshold degree of alignment with a pathogen protein sequence and have at least one insertion or deletion ("indel") relative to the pathogen protein sequence satisfying a predefined condition ; and b) identifying drug target suitability characteristics of a region of the pathogen protein sequence in response to the group of aligned sequences.

Identifying the characteristics of the aligned sequences may include, for each of the aligned sequences, identifying characteristics of the at least one inde and a complementarity region thereof.

Identifying characteristics of the at least one indel and the complemeniaiity region may include one or more of the following : identifying secondary structure of the

pathogen protein sequence and the aligned protein sequences ; identifying tertiary <BR> <BR> <BR> structure of the pathogen protein, sequence and five aligned protein sequences ; identifying quaternary structure of the pathogen protein sequence and the aligned protein sequences ; and d perforrning functional analysis of the pathogen protein.

The method may further include identifying a moiety capable of binding to a target region of the pathogen protein sequence.

In accordance with another aspect of the invention, there is provided a data structure embodied in a medium including respective'fields for storing: an identification of a protein sequence ; an identification of a pathogen protein sequence ; alignment data indicative of alignment identity between the protein sequence and the pathogen protein sequence ; and insertions or deletions ("indels") of the protein sequence relative to the pathogen protein sequence.

In accordance with another aspect of the invention, there is provided a data structure embodied in a medium for storing data produced by the methods disclosed herein, the data structure including respective fields for storing : an identification of the first protein sequence ; an identification of one of the first group of aligned protein sequences ; a. degree of alignment identity between the one of the first group and the first protein sequence ; and insertions or deletions of the one of the first group relative to the first protein sequence.

Embodiments of this invention may provide a method of identifying members of a group of homologous proteins, which members have a high probability of being affected by a drug or have a high probability of being bound to a specific binding moiety. In this specification, such homologous proteins are referred to as essential proteins. Members of a group of homologous proteins having or lacking an indel as defined herein may be identified or selected as targets. The indel or region lacking an indel may be used for design of a specific binding moiety as described herein. Embodiments of this invention may be particularly useful for identifying target homologous proteins from pathogenic organisms. Thus, such embodiments of this invention may involve indel comparison between species identified as either host or pathogenic.

Embodiments of this invention may also provide instructions, including instructions in machine-readable form. for operation of a. computer implemented method

for performing the aforementioned methods, including the specific embodiments of such methods described herein. Embodiments of this invention may also provide a machine- readable medium containing such. instructions. Embodiments of this invention may also provide a database in which information regarding indels is related or associated with information regarding homology of proteins in which the indels occur and which may also be related or associated'with information concerning whether a source organism is a host or a pathogen. Embodiments of this invention may also provide methods for using such a database for identification or selection of drug targets in pathogenic organisms and for providing specific sequence information which may be used for drug development or design. In the latter situation, such drug design may be based on three-dimensional models prepared by computer implemented methods, which models are based on the sequence information referred to above. Further, embodiments of this invention may provide methods in which protein fragments including indels or regions of homologous proteins not containing an indel are used for screening compounds capable of binding to or affecting the activity of such protein fragments.

Embodiments of this invention build upon the discovery that essential pathogen proteins can contain insertions or deletions compared to their host homologues conferring structural differences that may be exploited as drug targets specific for the pathogen.

Proteins that are essential to pathogen survival can contain. discrete, heretofore unrecognized indels which differentiate them from their otherwise highly related mammalian homologues. Up to now, because of their high degree of relatedness and because they are essential for survival in both host and pathogen, these proteins have not been candidate drug targets.

The importance of identification of relative insertions and deletions (or indels) in protein sequence ? has been previously described (e. g. PCT application publication no.

W097/01578). However, more have been no comprehensive indel investigations taking into account available sequence information and information concerning homologous proteins. Embodiments of this invention may provide a comprehensive database and method of analysis based on frequency of insertions and deletions amongst otherwise highly homologous proteins. Importantly, this type of analysis can identify a large data set of drug targets in pathogenic microorganisms. Once identified the unexpected

structural differences in otherwise homologous proteins may be exploited for targeted drug development.

Embodiments of this invention may permit the identification of pathogen proteins that are critical for fitness and survival and which also show insertions/deletions compared to their host homologues. A relative insertion or deletion in the sequence of an essential pathogen protein may confer to the region, a unique defined secondary structure, such as hairpin fold, and that such a structural feature may be exploited as potential site for selective drug binding. The information herein concerning EF-la is illustrative of the potential of certain embodiments of this invention and is not intended to be limiting as to the scope of this invention.

EF-la, or a close homologue, is expressed in all cells, whether they are eukaryotic or prokaryotic in origin. It serves as an essential enzyme for the translation of proteins from messenger RNA (mRNA) by controlling the speed and placement of tRNA-amino acid complexes in a growing polypeptide chain. Without EF-la, protein synthesis ceases, as does cell viability.

We have now characterized a pathogenic form of EF-la derived from protozoan sources and have found that the pathogenic form is a specific SHP-1 binding and activating protein. Protein modeling identifies important structural and functional differences between the pathogenic and mammalian forms of EF-la, with such differences being shared amongst other trypanosomatids which cause trypanosomiases, and unrelated protozoa. Further, infection of macrophages with leishmania causes redistribution of host as well as pathogen EF-la. The distinct properties of pathogen EF- la in comparison to the human protein and the essential role for EF-la in protein translation and other cell functions, makes EF-1 a a drug target for treatment of protozoan related diseases.

Molecular modeling of EF-la illustrates a species differentiated 12 amino acid 'indel' (Lnsertion/deletion sequence) that forms a surface exposed hairpin loop on human EF-la, which is conversely deleted in the protozoan protein. Thus, there is a significant region of differentially exposed amino acids between protozoan and human EF-la proteins, two proteins that are otherwise highly conserved. Based upon the results herein, and the known phylogeny of protozoan or protozoan like organisms, there are numerous pathogens that lack the aforementioned indel in EF-la, including: leishmania (agent of

leishmaniasis), Trypanosoma brucei and Trypanosoma cruzi (respectively agents of sleeping sickness and Chagas discase), cryptosporidium (agent of cryptosporidiosis), entamoeba (agent of amebiasis), Giardia (agent of giardiasis), plasmodium (agent of malaria), Tovoplasma gondi (aii agent of toxoplasma) and Pneumocystitis carinis (an agent of pneumocystitus). EF-1α is multifunctional atzd plays at least two essential roles, these being the regulation of protein synthesis and as a virulence factor. A virulence factor renders a host cell permissive for infection and may possess one or more of the following : a component of a pathogen that when deleted specifically impairs virulence but not viability ; a microbial product that permits a pathogen to cause disease ; a component of a pathogen that damages the host ; or an effector molecule that 1) produces changes in target cells outside of the organism itself, 2) is found only in pathogenic species or variety of the organism, and 3) is located on the surface of the organism or is secreted by the organism.

An example of homologous, essential proteins are the ER proteins firom any of the sources described herein. Preferably, such homologous proteins as referred to herein generally will have at least 70% sequence identity as determined by suitable algorithms, <BR> <BR> <BR> <BR> including BLAST using default parameters, Prefembly, sequence identity between homologous proteins will be high, with approximate percentage identity from least to most preferable being: 70, 75, 80, 85, 80 and greater than 90%.

The additional 12 amino acids found within human EF-1α form a hairpin-loop structure and is an indel. This indel is located on the exterior of the protein and is closely opposed to the main body of EF-la. The"unshielded"region of the pathogen EF-1α protein, which is concealed or shielded by the indel in the human protein (henceforth referred to as an"indel complementarity region"), contains several highly polar residues with exposed side chains. Therapeutic compounds can bind specifically to this indel complementarity region on pathogen EF-la and disrupt its function, without interfering with the corresponding region in the host homologue that is concealed by the indel. In addition, compounds that bind to pathogen. EF-1 a, but not to the human homologue can be used for effective diagnostic tests. Such binding compounds include antibodies, peptides, proteins and small molecule compounds.

Embodiments of this invention may include methods : Cor identifying, detecting, or assaying pathogenic proteins which include determining whether a protein member of a

homologous group of proteins includes an indel as described herein. These methods may be performed, for example, by sequencing (actual or predicted) and comparison of such sequence to homologous proteins, particularly those from pathogenic and host organisms.

Embodiments of this invention may include methods for identifying, detecting, or assaying moieties for specific binding to pathogenic proteins or an ability to modulate the activity of a pathogenic moiety, which methods include determining whether a putative specific binding moiety binds to a pathogenic protein as described herein and not to a host protein (or binds to the pathogenic protein to a greater extent than the host form); or, utilizing a first three-dimensional model of a pathogenic indel or indel complementarity region and a second three-dimensional model of a putative specific binding moiety, positioning both the three-dimensional models to form a third three-dimensional model of a specific binding complex and determining whether the binding complex is favourable (for example, without unacceptable hindrances such as steric, electrostatic, and hydrophobic hindrances). The latter methods may also be used for designing a specific binding moiety for pathogenic protein wherein a three-dimensional model of a putative specific binding moiety is altered to optimize binding between the moiety and the indel or indel complementarity region. Using the example herein, an indel complementarity region may include regions of contiguous amino acids from the following regions of leishma7çia EF-la or from corresponding (homologous) regions of another pathogenic EF-la : amino acids 215-224 TLLDALDMLE amino acids 186-194 EKVRFIPIS amino acids 158-168 KTVTYAQSRYD.

In the case of the example, the specific binding moiety may be capable of binding with, or such binding may be optimized to occur with one or more contact residues such as any of the following amino acids from leishma71ia EF-la or a similarly placed charged or polar amino acid from another pathogenic EF-la : Glutamate (E) 186 Lysine (K) 187 Arginine (R) 189 Aspartate (D) 218 Aspartate (D) 221

Methionin (M) 222 Glutamate (E) 224 The aforementioned methods may additionally include one or more of : synthesizing a specific binding moiety, and combining a specific binding moiety with a pathogenic protein. These methods may further include methods of detecting, identifying, or assaying binding to the protein, as described above.

Embodiments of this invention may also include moieties capable of specific binding to pathogenic proteins identified, detected, synthesized, designed or assayed by the methods of this invention. By"specific binding"it is meant that the moiety is capable of preferentially binding to a pathogenic protein as described herein as compared to host proteins. Preferably, such a moiety will exhibit substantially no binding to the host form.

Such binding moieties may be employed in assays for the protein and as specific targeting ligands (e. g. for labeling or directing a further moiety to the location of the pathogenic protein). In some embodiments, such moieties may modulate the activity of the protein.

Preferably, the latter moieties will modulate the activity of pathogenic proteins in terms of essential function or function as a pathogen virulence factor. Specific binding moieties of embodiments of this invention may be chemical moieties designed to bind with contact residues in an indel or indel complementarity region as defined herein. Such moieties may be peptides or non-peptide molecules. Such moieties may be antibodies or antibody fragments capable of binding to an antigenic determinant of an indel or indel complementarity region. In addition, specific binding moieties of embodiments of this invention may shield the indel complementarity region without directly binding to the region. Specific binding moieties of embodiments of this invention may include therapeutic moieties for treatment of infections by organisms possessing pathogenic proteins. Accordingly, embodiments of this invention may include the use of such therapeutic moieties in treatment and may provide pharmaceutical compositions and formulations including specific binding moieties of embodiments of this invention together with one or more acceptable pharmaceutical carriers.

Embodiments of this invention may also include a method for modulating the activity of a pathogenic protein described herein including contacting a pathogenic protein with a specific binding moiety of an embodiment of this invention capable of modulating a function of the protein. Preferably, the modulation includes antagonism of a pathogenic function.

BRIEF DESCRIPTION OF THE DRAWINGS Figure 1 is a block diagram of a system for protein sequence analysis according to a first embodiment of the invention.

Figure 2 is a flow chart of a protein sequence analysis routine executed by a processor circuit of the system shown in Figure 1.

Figure 3 is a block diagram of a system for protein sequence analysis according to a second embodiment of the invention.

Figure 4 is a flow chart of a protein partition routine executed by at least one processor circuit of the system shown in Figure 3.

Figure 5 is a flow chart of a BLAST analysis execution routine executed by at least one processor circuit of the system shown in Figure 3.

Figure 6 is a flow chart of a BLAST analysis distribution routine executed by at least one processor circuit of the system shown in Figure 3.

Figure 7 is a flow chart of a significant indel identification routine executed by at least one processor circuit of the system shown in Figure 3.

Figure 8 is a flow chart of a significant indel analysis distribution routine executed by at least one processor circuit of the system shown in Figure 3.

Figure 9 is an illustrative example of a data structure produced by the system shown in Figure 3.

DETAILED DESCRIPTION Definitions In this specification, the term"amino acid"means any of the organic compounds that contain one or more basic amino groups and one or more acidic carboxyl groups and that are polymerized to form peptides and proteins; only 20 of the more than 80 amino acids found in nature serve as building blocks for proteins.

In this specification, the term'protein sequence"means a representation of the amino acid constituents that make up a protein or peptide. A protein sequence is usually represented by a linear sequence of the standard single letter codes for the constituent amino acids.

In this specification, the term"essential"with reference to a protein, refers to the product of a gene where a mutation in the gene or the absence of the gene product or the absence of function of the gene product is detrimental to the pathogen. Whereby, detrimental to the pathogen means that such a mutation in the gene or the absence of the gene product or the absence of function of the gene product will result in death of the pathogen, lack of pathogen viability, lack of pathogenicity of the pathogen, and/or inability of the pathogen to reproduce in the host organism.

In this specification, the term"indel"refers to a region of amino acid sequence that includes all insertions and deletions in at least one amino acid sequence when aligned with another amino acid sequence, where the two sequences share overall similar sequences of amino acids. Where two sequences have overall similar sequences, that means that the two sequences when aligned contain primarily identical or conserved amino acids at corresponding amino acid positions. Typically, such an indel will have a maximum number of amino acids the same as or less than the following: 50,40, 30,25, 20 and 15. Also the number of amino acids in an indel needed to be considered significant is any number greater than three.

In this specification, the term"pathogen"means any organism (usually a microorganism) which by direct interaction with (infection of) a second organism (by convention a

multicellular organism) causes disease in the second organism. The definition as used herein is also meant to include parasites and toxinogenic organisms.

In this specification, the term"host"means any organism on or in which a pathogen or parasite or toxinogenic organism is able to live or reproduce.

In this specification, the term"identity"means that all the amino acids at a single position are identical when two or more amino acid or protein sequences are aligned.

Methods A) Protein Sequence Analysis Method-Indel Identification To perform a comprehensive indel analysis, sequences may be collected for analysis from numerous databases, as represented by the following examples : SWISS-PROT a protein sequence database that also provides a high level of annotations relating to the function of a protein, its domains structure, post- translational modifications, variants, etc. (Bairoch A. and Apweiler R. (2000) Nucleic Acids Res. 28 (1) : 45-48; Bairoch A. and Apweiler R. (1997) J. Mol. Med.

75 (5): 312-316; Junker. L. et al. (1999) Bioinformatics 15 (12): 1066-1007) ; TrEMBL a computer-annotated supplement of SWISS-PROT that contains all the translations of EMBL nucleotide sequence entries (Bairoch A. and Apweiler R.

(2000) Nucleic Acids Res. 28 (1) : 45-48); and nr database compares all non-redundant GenBank CDS translations plus protein sequences from other databases such as PDB, SwissProt, PIR and PRF.

Alignments of protein sequences may be conducted using existing algorithms to search databases for sequences similar to a query sequence. Protein sequence alignments are not generally limited by protein type or organism.

One alignment method is the Smith-Waterman algorithm (Smith, T. F. and Waterman, M. S. 1981. Journal of Molecular Biology 147 (1) : 195-197), which uses dynamic programming to determine how an optimal alignment between the query

sequence and a database sequence can be produced. This alignment is obtained by determining what transformations the query sequence would need to undergo to match the database sequence. Transformations include substituting one character for another and inserting or deleting a string of characters. A score is assigned for each character-to- character comparison-positive scores for exact matches and some substitutions, negative scores for other substitutions and insertions/deletions. The first character in an insertion or deletion gap is scored with a gap open penalty and subsequent characters are scored with a gap extension penalty. Scores are obtained from statistically-derived scoring matrices. The combination of transformations that results in the highest score is used to generate an alignment between the query sequence and database sequence.

The Needleman-Wunsch (Needleman, S. B. and Wunsch, C. D. 1970. Journal of Molecular Biology 48 (3): 443-453) algorithm is similar to the Smith-Waterman algorithm, but sequence comparisons are global, not local. Global comparisons force an alignment of the entire query sequence against the entire database sequence. While local alignments always begin and end with a match, global alignments may begin or end with an insertion or deletion (indel). For a given query sequence and database sequence, a global score will be less than or equal to a local score due to indels on the ends.

As an alternative to the above algorithms, a Hidden Markov Model (HMM) search (Eddy, S. R. 1996. Current Opinion in Structural Biology 6 (3): 361-365) could be used to generate protein sequence alignments. HMM scoring weighs the probability of a match being followed by insertions/deletions or vice-versa. In addition, HMMs allow insertion to deletion transitions (and vice versa) and scoring of begin and end states to control whether a search is run globally or locally.

One or more of the above algorithms may be used in an alignment program to generate protein sequence alignments. A person skilled in the art has numerous sequence alignment programs to choose from, that incorporate a variety of different algorithms.

One example of an alignment program is BLASTP (Altschul, S. F., et al. (1997) Nucleic Acids Res. 25 (17): 3389-3402). Other alignment programs are CLUSTAL W and PILEUP. The standard output from a BLASTP run contains enough information to conduct further indel analysis as described below.

The alignment results are collected in an output file, where each individual sequence is identified by protein, organism, family, genus, superkingdom and

characterized by whether the organism is parasitic or free living, pathogenic or non- pathogenic etc.

The alignment results that are collected are then analyzed (parsed) to identify potentially significant indels. From the initial alignment results, alignments are identified which share at least 70% identity and preferably greater than 70% identity. Only those that have at least 70% identity are further selected from.

When aligned, the sequences produce gaps, which are predicted indels. These predicted indels may then be characterized on the basis of their size (e. g. four or more amino acids). The sequence identity and indel size information associated with each sequence alignment is then stored for subsequent retrieval and analysis.

Exemplary System Referring to Figure 1, an exemplary apparatus for protein sequence analysis according to a first embodiment of the invention is shown generally at 20. In this embodiment, the apparatus includes a processor circuit 21, which is configured to identify, from among a first group of aligned protein sequences having at least a threshold degree of alignment with a first protein sequence, a second group of at least one of said aligned sequences having at least one insertion or deletion ("indel") relative to said first protein sequence satisfying a predefined condition.

In this embodiment, the processor circuit 21 includes a microprocessor, which may be housed in a general purpose or special purpose computer (not shown), for example. More generally, however, in this specification, the term"processor circuit"is intended to broadly encompass any type of device or combination of devices capable of performing the functions described herein, including (without limitation) other types of . microprocessors, microcontrollers, other integrated circuits, other types of circuits or combinations of circuits, logic gates or gate arrays, or programmable devices of any sort, for example, either alone or in combination with other such devices located at the same location or remotely from each other, for example. Additional types of processor circuits will be apparent to those ordinarily skilled in the art upon review of this specification, and substitution of any such other types of processor circuits is considered not to depart from the scope of the present invention as defined by the claims appended hereto.

In this embodiment, the processor circuit 21 is in communication with a random access memory (RAM) 22, which may be either separate from or integral with the

processor circuit, or which may include a combination of on-board and external RAM.

The processor circuit 21 is also in communication with a storage medium 23, which in this embodiment includes a hard disk drive, although alternatively, other types of storage media may be substituted. In this embodiment, the storage medium 23 acts as a computer-readable medium storing various codes for directing the processor circuit 21 to carry out the functions disclosed herein, including an analysis routine 24. Alternatively, however, such codes may be provided by other computer-readable media. For example, fixed media such as a compact disc or floppy diskette, or a transmission medium such as a communications network, may provide such codes. Generally, any medium capable of providing signals for directing the processor circuit 21 to perform the functions disclosed herein may be substituted if desired. In this embodiment, the storage medium 23 also includes a results store 25, for storing data structures including results of the execution of the analysis routine 24 by the processor circuit 21.

In this embodiment, the processor circuit 21 is also in communication with a sequence database 26, which may include may include any one or more of the SWISS- PROT, TrEMBL or nr databases referred to above, or any other suitable sequence database. The processor circuit 21 may be in direct communication with the database, or may be in indirect communication therewith, for example, via a network (not shown). If desired, the storage medium 23 may store the desired sequences and may thus act as the sequence database 26.

In this embodiment, the processor circuit 21 is also in communication with a user interface 27, which may include a keyboard and a display, for example.

Referring to Figures 1 and 2, the analysis routine is shown generally at 24 in Figure 2. Generally, the analysis routine 24 configures the processor circuit 21 to identify, from among a first group of aligned protein sequences having at least a threshold degree of alignment with a first protein sequence, a second group of at least one of said aligned sequences having at least one insertion or deletion ("indel") relative to said first protein sequence satisfying a predefined condition. More particularly, in this embodiment the first protein sequence includes a protein sequence of a pathogen, although alternatively, other types of protein sequences may be substituted if desired.

The identification of the first protein sequence may be supplied by a user (not shown) of the apparatus 20 at the time the analysis routine 2 is first executed. For example, the user

may use the user interface 27 to either manually enter the first protein sequence or to enter a location or filename accessible by the processor circuit in which the first protein sequence is stored. The processor circuit 21 may store the first protein sequence in either the RAM 22 or the storage medium 23 or both, for use in execution of the analysis routine.

The analysis routine 24 includes a first block 200 of codes, which directs the processor circuit 21 to identify a first group of aligned protein sequences, or in other words, protein sequences having a substantial alignment identity with the first protein sequence. To achieve this, block 200 directs the processor circuit 21 to execute the BLASTP routine referred to earlier herein, to identify aligned protein sequences stored in the database 26, i. e. , protein sequences that are aligned with the first protein sequence.

Block 200 directs the processor circuit to store the output results of the BLASTP routine in an alignment results store 201 of the results store 25 of the storage medium 23.

Alternatively, however, any other suitable alignment routine may be substituted for the BLASTP routine executed at block 200.

Block 202 then directs the processor circuit 21 to address the next successive aligned protein sequence stored in the alignment results store 201 in the storage medium 23, commencing with the first such aligned sequence when block 202 is first executed.

Block 204 then directs the processor circuit 21 to determine whether the degree of alignment, or in other words the alignment identity, between the currently addressed aligned protein sequence and the first protein sequence, is at least equal to a threshold degree of alignment. In this embodiment, the threshold degree of alignment is 70%.

Alternatively, however, depending upon the application, other threshold degrees of alignment may be substituted. For example, minimum threshold degrees of alignment identity of 75%, 80%, 85%, or 90% may be preferred, depending upon the nature and scope of the particular application. Preferably, the threshold degree of alignment is an adjustable parameter of the analysis routine 24, that the user may adjust prior to execution of the analysis routine by entering the desired threshold using the user interface 27, for example. If the alignment identity between the currently addressed protein sequence alignment and the first protein sequence is less than the threshold degree, then the currently addressed protein sequence is effectively discarded as irrelevant, and the

processor circuit is directed to block 210 below to determine whether any further aligned sequences remain to be analyzed.

If at block 204 the alignment identity between the currently addressed aligned protein sequence and the first protein sequence is determined to be greater than or equal to the threshold degree of identity, block 206 directs the processor circuit 21 to determine whether the currently addressed aligned protein sequence has at least one indel relative to the first protein sequence satisfying a predefined condition. In this embodiment, the predefined condition includes a minimum length of the indel. More particularly, in this embodiment the minimum length of the indel is at least four contiguous amino acids. In this embodiment, the predefined condition also includes a maximum length of the indel.

More particularly, in this embodiment the maximum length is 50 contiguous amino acids.

Effectively, therefore, in this embodiment the predefined condition includes a range of length of the at least one indel, which in this embodiment is a range of four to 50 contiguous amino acids. Alternatively, however, other types of predefined conditions may be substituted, depending upon the nature and scope of the particular application.

Preferably, the predefined condition is an adjustable parameter of the analysis routine 24, that the user may adjust prior to execution of the analysis routine by entering the desired threshold using the user interface 27, for example. If the currently addressed aligned protein sequence does not have at least one indel satisfying the predefined condition, then the currently addressed protein sequence is effectively discarded as irrelevant, and the processor circuit is directed to block 210 below to determine whether any further aligned protein sequences remain to be analyzed.

If at block 206 it is determined that the currently addressed aligned protein sequence does have at least one indel satisfying the predefined condition, block 208 directs the processor circuit 21 to store an identification of the currently addressed aligned protein sequence. In this embodiment the processor circuit is directed to store the addressed sequence in an alignments store 209 of the results area 25 of the storage medium 23, although alternatively, the aligned protein sequence may be stored elsewhere, such as the RAM 22, for example.

Block 210 then directs the processor circuit 21 to determine whether all of the aligned protein sequence results produced at block 200 have been addressed at blocks 202 to 208 above, and if not, the processor circuit is directed back to block 202 to address the

next aligned protein sequence in the alignment results store 201 and to proceed to execute blocks 204,206 and 20S as described above. It will be appreciated that the successive execution of block 204 on each of the aligned protein sequences produced at block 200 effectively directs the processor circuit to compare a plurality of protein sequences To the first protein sequence to identify the first group of aligned protein sequences having at least the threshold degree of alignment with the iirst protein sequence. Alternatively, this comparison, including the use of the threshold, may be combined into block 200 as discussed above, if desired.

If no more aligned sequences remain to be addressed, block 212 directs the processor circuit 21 to store the significant indels, i. e. , the second, group of aligned sequences ident ed at block 206 ; in the storage medium 23. To achieve this, in the present embodiment block 212 directs the processor circuit to Store the contents of the significant indels store 209 in a data structure within a text file In the storage medium 23.

Alternatively ? other storage methods or formats may be substituted. The analysis routine 24 is then ended.

If desired, the analysis routine 24 may be repeatedly executed to identify potentially signitlcant indels for a number of different pathogens or other specified first protein sequences. In other words, the processor circuit 21 may be configured to successively select each one of a plurality of protein sequences as the first protein sequence, and may be configured to repeat the identifying steps discussed above at blocks 204 and 206 for each one of the plurality of protein sequences. To achieve this, the analysis routine 24 may include additional blocks of codes 214 and 216 shown in brolcen outline in Figure 2. Block 214 directs the processor circuit 21 to determine whether any further pathogens or other first protein sequences have been specified by the user for analysis, and if so, block 216 directs the processor circuit to address the next such first protein sequence, and to return to block 200 to commence re-execution of the analysis routine 24 as discussed above.

Distributed Processing System Depending upon the scope of the required analysis and upon available processing power, the analysis of large numbers of proteins may be very time consuming and may

require significant computing resources if executed entirely on a single processor circuit.

As some applications of specific embodiments of the invention may entail analysis of large groups of protein sequences (for example, some databases include 800, 000 or more relevant protein sequences), the concurrent or parallel use of a plurality of processor circuits may'facilitate and expedite such analysis. Where a plurality of processor circuits are being used to analyze a subset of proteins it is possible to minimize data loss and time loss, because a latal error in a routine being executed on any one of the processor circuits handling an analysis job on a subset of proteins is likely to result in a smaller loss of data and time.

Accordingly, referring to Figure 3, an apparatus for protein sequence analysis according to a second embodiment of the invention is shown generally at 30. In this embodiment, the apparatus includes a processor circuit 32 in communication with a storage medium 33, a random access memory (RAM) 34, a media interface 35, and a user interface 37.

In the present embodiment the processor circuit 32 includes a microprocessor, Alternatively, however, other types of processor circuits may be substituted, such as those noted above in connection with the previous embodiment, ibr example.

Generally, the media interface 35 is used to read data signals from a computer- readable medium 36. More particularly, in this embodiment the media interface 35 includes a CD-RW drive for reading signals from the medium 36, which in this embodiment is a compact disc. Alternatively, any other suitable types of media interfaces and media may bs substituted. In this embodiment, the user interface 37 includes a keyboard and a display device (neither of which is shown in Fig. 3) In this embodiment, the processor circuit 32 is in communication with a network 40, which in this embodiment includes a local area network in communication with the Internet. The processor circuit 32 is in communication, via the network 40, with additional processor circuits such as those shown at 42, 44, 46, 48, 50 and 52, for example. In this embodiment, each of the processor circuits 32, 42, 44, 46, 48, 50 and 52 is in communication via the network 40 with a sequence database 38, which may include any one or more of the SWISS-PROT, TrEMBL or nr databases referred to above, or any other suitable sequence database.

In this embodiment the storage medium 33 includes a hard disk drive, although other types of storage media may be substituted if desired. In the present embodiment, the storage medium 33 acts as a computer-readable medium storing codes for directing one or more of the processor circuits shown in Figure 3 to perform the various functions disclosed herein. More particularly, in this embodiment the storage medium 33 stores a plurality of analysis routines 100, including a protein partition routine 300, a BLAST execution routine 400, a BLAST distribution routine 500, an indel identification routine 600, and an indel distribution routine 700. Alternatively, such routines may be stored on other computer-readable media, such as the medium 36, for example. In addition, in this embodiment the network 40 acts as a transmission medium providing codes for directing the processor circuits 42 to 52 to perform the various functions discussed below. More generally, any alternative way of generating data signals for directing the various processor circuits to perform the functions disclosed herein may be substituted.

Referring to Figures 3 and 4, the protein partition routine is shown generally at 300 in Figure 4. The protein partition routine 300 begins with a first block of codes 302 that directs the processor circuit 32 to access the sequence database 38.

Block 304 then directs the processor circuit 32 to extract a set of 1000 proteins and block 306 directs the processor circuit 32 to assign a name to the protein set according to the order that the protein set was extracted from the protein database (i. e. the fifth set would be"proteinset5k. txt").

Block 308 directs the processor circuit 32 to assign a directory to the protein set (or to create one if one does not exist) In this embodiment, each directory includes 100 sets of proteins, each set including 1000 individual proteins. For example, the first directory is"/proteinsets/protset100k/").

Block 310 directs the processor circuit 32 to store each set of proteins in a file with an assigned filename and place the file in a protein set directory in the storage medium 33. Block 312 then directs the processor circuit 32 to create a job file associated with the set (for example, set 5 would be assigned to job file"job_5k. txt"). Such job files are effectively used for communicating tasks to the various distributed processor circuits, as discussed in greater detail below. Block 312 directs the processor circuit 32 to store the job file in a job file directory 313 in the storage medium 33.

Finally, block 314 directs the processor circuit 32to return to block 302 to select another set of proteins from the protein database and assign a filename, directory and job file to the selected set as described above until no more sets are found in the protein database.

Referring to Figures 3 and 5, the BLAST analysis execution, routine is shown generally at 400 in Figure 5. Generally, in this embodiment, the BLAST analysis execution routine 400 directs the processor circuit to obtain a job file created under the direction of the protein partition routine 300 and to execute an alignment algorithm, which in this embodiment is a BLAST analysis, on the contents of the job 61e. Such job Eiles are eNectively distributed under the direction of the BLAST analysis distribution routine 500, discussed below. Thus, in view of the latter distribution of job files, the BLAST analysis execution routine 400 may be executed on any one of the processor circuits 32, 42, 44, 46, 48, 50 or 52 shown in Figure 3. For ease of understanding, therefore, the present routine is described as being executed by the"processor circuit", with the understanding that any one of the distributed. processor circuits may execute this routine.

The BLAST analysis execution routine 400 begins with a first block of codes 402, which directs the processor circuit to obtain a job file (created by the processor circuit 32 under the direction of the protein partition routine 300, as discussed above) from the storage medium 33.

Block 404 then directs the processor circuit to move the job file from the job file directory 313 to a running job directory 405. This ensures that any fatal crashes by the processor circuit running the BLAST analysis execution routine 400 can easily be identified and allows for rc-analysis if the BLAST analysis was not complete. Block 404 also directs the processor circuit to store an identification of itself in association with the file stored in the running job directory, to allow the processor circuit 32 to identify the processor circuit executing the particular job identified by the job file stored in the running job directory.

Block 406 directs the processor circuit to locate the protein set in the protein set directory 311 relating to the job file being analyzed.

Block 408 then directs the processor circuit to execute an alignment algorithm, which in this embodiment is a BLAST analysis routine as described above, on the protein

set. Alternatively, other suitable alignment algorithms or methods may be substituted. It will be recalled that the protein set of the job file in the present embodiment includes 1000 protein sequences. Accordingly, in this embodiment, the alignment algorithm is effectively executed 1000 times in succession, each execution employing a different respective one of the 1000 protein sequences as the query sequence. With each such execution corresponding to each such selected query sequence, the processor circuit is directed to search the sequence database 38 for aligned protein sequences, i. e. , protein sequences having a substantial alignment identity with the query sequence. The processor circuit is directed to store each such located set of aligned protein sequences in association with an identification of the query sequence that was used to locate the aligned sequences, as a query set. The processor circuit is directed to continue selecting successive aligned protein sequences of the protein set as the next query sequence, to repeat such searching of the database for sequences aligned with the query sequence, and to continue repeating such searching for each successive query sequence until all of the 1000 sequences of the protein set of the job file have been used as query sequences.

Block 410 then directs the processor circuit to store the completed BLAST results as a compressed text file (using gzip) to save disk space in a"local/tmp/directory"of a local storage medium accessible by the processor circuit. It will be appreciated that this BLAST results file effectively includes 1000 respective query sets representing the results of 1000 successive alignment searches for sequences aligning with 1000 respective query sequences.

Block 420 then directs the processor circuit to create a"doneblast"or BLAST output text file in the local/tmp/directory to effectively notify the processor circuit 32 that the BLAST analysis is complete (as discussed below in connection with the BLAST distribution routine 500, block 510 of the latter routine directs the processor circuit to periodically monitor the contents of the local/tmp/directory areas of the distributed processor circuits, to determine whether each such processor circuit has completed a BLAST analysis job). Block 430 then directs the processor circuit to log any errors that were encountered when running the BLAST analysis execution routine 400, in a local error log (not shown).

Referring to Figures 3 and 6, the BLAST analysis distribution routine is shown generally at 500 in Figure 6. Generally, the BLAST analysis distribution routine 500

configures the processor circuit 32 to assign jobs to the various other processor circuits, each of which may execute a BLAST analysis execution routine 400 as described above.

In this embodiment, the BLAST analysis distribution routine 500 begins with a first block of codes 502 that directs the processor circuit 32 to retrieve a list of job files from the job files directory 313 in the storage medium 33.

Block 504 then directs the processor circuit 32 to determine which of the various distributed processor circuits shown in Figure 3 are available for execution of BLAST analysis. This may be achieved by directing the processor circuit to communicate directly with the various distributed processor circuits, or alternatively, by communicating with a central network node, for example. In this embodiment, a processor circuit is considered to be available for BLAST analysis if it has a MIPS S > 2000/equal to or better than a dual 1 G CPU and load averages of less than 2, although alternatively, other availability criteria may be substituted.

Block 506 then directs the processor circuit 32 to distribute one BLAST job to each of the available processor circuits. In this embodiment, the processor circuit 32 signals each such available processor circuit to execute the BLAST analysis execution <BR> <BR> routine 400, to obtain and analyze a job file as described above. Alternatively, any other suitable way of distributing tasks may be substituted.

Blocks 508 to 518 effectively direct the processor circuit 32 to monitor all BLAST jobs running on the available processor circuits. Block 508 directs the processor circuit to address the next running job identified in the running jobs directory 405 shown in Figure 3, corxanencing with the first such running job when block 508 is first executed.

Block 510 directs the processor circuit 32 to determine whether the currently addressed running job has failed. This may be achieved, for example, by directing the processor circuit 32 to examine the contents of the local error log of the processor circuit in question (identified by the contents of the addressed running job directory record), to determine whether the error log includes an indication of a fatal error (as discussed above in connection with block 430 of the BLAST execution routine). This may also be achieved by determining whether a timeout has occurred, i. e., whether the processor circuit has been executing the running job for a longer period of time man expected without having completed it. Alternatively, other ways of detecting job failure may be substituted.

If at block 510 a job failure is detected, block 512 directs the processor circuit 32 to move the addressed job ille Irom the running jobs directory 405 back to the job files directory 313, so that the job may be subsequently executed by another available processor circuit. The processor circuit is then directed to block 5 8 below.

If no job re is detected, block 514 directs the processor circuit 32 to determine whether the addressed job has been completed. To achieve this, block 514 directs the processor circuit 32 to examine the contents of the e local/tmp/directory of the processor circuit associated with the addressed job, to determine whether a BLAST output text file has been stored therein, as discussed above in connection with block 420 of the BLAST execution routine.

If at block 514 it is determined that the j ob was successfully completed, block 516 directs the processor circuit 32 to compress and move the BLAST output file from the local/inp/directory on the available processor circuit running the BLAST analysis execution routine 400 to a completed BLAST results directory 515 in the storage medium 33.

Following execution of block 516 or block 512, or if at block 514 it was determined that the job is incomplete (but not failed at block 510), then block 518 directs the processor circuit 32 to determine whether each of the jobs listed in the running jobs directory 405 has been addressed at blocks 508 through 516 above. If not, the processor circuit is directed back to block 508 to address the next successive running job record as discussed above.

If all running jobs have been addressed, block 520 directs the processor circuit 32 to determine whether the job files directory 313 contains any job files (i. e. , jobs which have not yet been executed). If so, the processor circuit 32 is directed back to block 502 to continue distributing jobs among the various processor circuits as discussed above.

Referring to Figures 3 and 7, the significant indel identiEcation routine is shown generally at 600 in Figure 7. Generally, the significant indel identification routine con-figures any one of the available processor circuits shown in Figure 3 to further analyze the BLAST alignment results (stored as BLAST output files) generated by the BLAST execution routine 400 a. ud stored in me BLAST results directory 515 as discussed above in connection with block 516 of the BLAST distribution routine 500-In this embodiment,

such further analysis of the BLAST output file (BLAST alignment results) involves identification of what are termed in this embodiment as siyificant indels. Significance" of an indel is determined by pre-defined criteria, which in this embodiment include percentage of alignment identity and the number of consecutive amino acids comprising an indel.

Under the direction of the indel distribution routine 700 discussed below, the processor circuit 32 may effectively assign indel analysis tasks to each of the available processor circuits shown in Figure 3, to cause such available processor circuits to execute the significant indel identification routine. For ease of understanding, therefore, the present routine is described as being executed by the "processor circuit", with the understanding that any one of the distributed processor circuits may execute this routine.

The significant indel identification routine 600 begins wu a first block of codes 602, which directs the processor circuit to retrieve a BLAST output file stored in the blast output directory 515 in the storage medium 33.

Block 603 directs the processor circuit to address the next query set stored in the retrieved BLAST output file, commencing with the first such query set when block 603 is first executed. In this regard, it will be recalled that in the present embodiment, the BLAST output file includes 1000 such query sets, each query set including an identification of the query sequence, and identifications of aligned protein sequences having a substantial alignment identity with the query sequence, as located by the BLAST algorithm.

Block 604 then directs the processor circuit to address the next aligned protein sequence stored in the currently addressed query set of the retrieved BLAST outptu file, commencing with the first such aligned sequence when block 604 is first executed, Block 606 then directs the processor circuit to determine whether the addressed aligned sequence has an alignment identity of greater than or equal to a threshold degree of alignment (which in this embodiment is 70%) with the query sequence of the currently addressed query set. If the addressed aligned sequence has less than the threshold degree of alignment, the processor circuit is effectively directed to discard the addressed aligned sequence, and is directed to block 617 below.

If the addressed aligned sequence has at least the threshold degree of alignment with the query sequence of the addressed query set, block 608 directs the processor circuit

to determine whether the currently addressed aligned sequence has at least six gaps relative to the query sequence, each gap representing a single amino acid of the query sequence that has no corresponding amino acid at the corresponding location in the currently addressed aligned sequence, or vice versa. In this embodiment, this may be achieved by determining whether the"gap value"in the BLAST output file for the aligned sequence is greater than or equal to 6 (6 amino acid gaps anywhere in either protein sequence). If not, the processor circuit is effectively directed to discard the aligned sequence and is directed to block 617 below.

If the inquiry at block 608 is answered in the affirmative, blocks 610 and 612 direct the processor circuit to evaluate whether there is a significant indel in the aligned sequence. To achieve this, block 610 directs the processor circuit to concatenate the alignment into a single line of sequence, then block 612 directs the processor circuit to determine whether the currently addressed aligned sequence has at least one indel satisfying a predefined condition. In this embodiment, the predefined condition is a minimum length of the indel, which in this embodiment is a length of at least six contiguous amino acids. Alternatively, other suitable predefined conditions may be substituted, such as those discussed above in connection with the previous embodiment, for example. If the addressed aligned sequence does not have at least one indel satisfying the predefined condition, the processor circuit is effectively directed to discard the aligned sequence and to proceed to block 617 below.

Conversely, if the aligned sequence has at least one indel satisfying the predefined condition, block 614 directs the processor circuit to obtain additional information about the currently addressed query sequence and aligned sequence. To achieve this, block 614 directs the processor circuit to execute a routine called"getz", which obtains additional information about the pair of aligned protein sequences (such as organism name and protein type). Such information may be achieved by directing the processor circuit to query the sequence database 38, for example, although alternatively, other ways of obtaining such information may be substituted.

Block 616 then directs the processor circuit to locally store ("local/temp/directory") the additional information along with the corresponding alignment information in a local storage medium accessible by the processor circuit.

Block 617 then. directs the processor circuit to determine whether the currently addressed query set includes any additional aligned protein sequences that have not yet been addressed at blocks 604 er seq. above, and if so, the processor circuit is directed back to block 604 to address the next successive aligned protein sequence of the currently addressed query set.

If no further aligned protein sequences in the currently addressed query set remain to be addressed, block 618 directs the processor circuit to determine whether the retrieved BLAST output file includes any additional query sets corresponding to additional respective query sequences, which have not yet been addressed at blocks 603 et seq. above. If so, the processor circuit is directed back to block 603 to address the next successive query set in the BLAST output file.

If no further query sets remain to be addressed, block 620 directs the processor circuit to store all of the alignment information stored above at block 616, in a consolidated text file, for subsequent retrieval by the processor circuit 32 shown in Figure 3.

Referring to Figures 3 and 8, the significant indel analysis distribution routine is shown generally at 700 in Figure 8. Generally, the significant indel analysis distribution routine directs the processor circuit 32 to distribute indel analysis jobs to any of the various available processor circuits shown in Figure 3, fox distributed execution of the significant indel identification routine 600 discussed above.

The significant indel analysis distribution routine 700 begins with a first block of codes 702 that directs the processor circuit 32 to retrieve a list of job files from BLAST results directory 515.

Block 704 then directs the processor circuit 32 to determine which of the various distributed processor circuits shown in Figure 3 are available for execution of the indel identification routine 600. This may be achieved by directing the processor circuit to communicate directly with the various distributed processor circuits, or alternatively, by communicating with a central network node, for example. In this embodiment, a processor circuit is considered to be available if it has a MIPS# 2000/ equal to or better than a dual 1 GHz CPU and load averages of less than 2, although alternatively, other availa. bility criteria may be substituted.

Block 706 then directs the processor circuit 32 to effectively distribute one significant indel identification job to each of the available processor circuits. Such distribution may occur in a manner similar to that described above in connection with block 506, for example, and upon distribution of an indel identification job to a processor circuit, the identification of that job is removed from the BLAST output directory 515.

Block 708 further directs the processor circuit 32 to monitor all significant indel identification jobs running on the available processor circuits. In this regard, block 708 directs the processor circuit 32 to check each individual available processor circuit to see if the significant indel identification routine (length polymorphism program) 600 for a given job was successfully executed. This determination may be achieved in a manner similar to that discussed above in connection with block 514, for example. If the job was not successfully completed, block 708 directs the processor circuit 32 to move the job back to the BLAST results directory 515 and clear the failed significant indel identification analysis job from the processor circuit in question. However, if the job was successfully executed, then block 708 directs the processor circuit 32 to move the stored indel analysis results from the local/tmp/directory on the available processor circuit to a completed significant indel analysis directory 713 in the storage medium 33.

Block 714 then directs the processor circuit 32 to determine whether the BLAST results directory 515 includes any further indel analysis jobs that have not yet been executed, and if so, the processor circuit is directed back to block 702 to continue distributing indel analysis jobs until all such jobs have been completed.

Referring to Figures 3 and 9, a partial data structure produced by the processor circuit under the direction of the significant indel analysis routine 600 is shown generally at 900 in Figure 9. In this embodiment, the data structure 900 includes respective fields, including a field 902 for storing an identification of the first protein sequence; a field 904 for storing an identification of one of the first group of aligned protein sequences having at least the threshold degree of alignment with the first sequence and having at least one indel relative to the first sequence satisfying the predefined condition; a field 906 for storing a degree of alignment identity between the one of the first group of aligned sequences and the first sequence; and field 908 for storing indications of insertions or deletions of the one of the first group of aligned sequences relative to the first protein sequence. In at least some cases, the first protein sequence will represent a pathogen

protein sequence, and therefore, in such cases, the field 902 stores an identification of a pathogen protein sequence.

As an example of the above mentioned method, significant indels have been identified in elongation factor-la (EF-la) from trypanosomatids and other protozoa.

The results of this example are described in further detail below.

B) Protein Sequence Analysis Method-Identification of Drug Target Suitability Characteristics Sequences with an alignment identity of > 70% and with gaps of four amino acids or greater may be stored together in a database for further structural analysis, for example to determine the structure and to determine frequency and type of amino acids surrounding predicted indels and alternatively to look at the region of the pathogen protein"unshielded", where the corresponding human protein is concealed or shielded by the indel (henceforth referred to as an"indel complementarity region"). The results of the alignments and subsequent indel analysis would also facilitate further functional and experimental analysis. Pathogen proteins may be further characterized to determine their role in the pathogens life in the host organism and importance for pathogen survival, reproduction or ability to cause disease in the host organism and therefore to decide whether or not the protein is a potential target. If the functional analysis reveals that the pathogen protein is"essential"whereby in the absence of function of the protein will result in death, lack of viability, lack of pathogenicity, or incapability of reproduction of the pathogen in the host organism then it may be termed an"essential"protein.

Alignments that satisfy the above criteria can be stored with their accompanying information in a database or data structure to be retrieved and further analyzed later.

The indel sequence, the indel complementarity region and overall protein sequence can be further analyzed based on secondary structure (PHD program-Rost, B. and Sander. C. (1993) J. Mol. Biol. 232 (2): 584-599 ; Rost, B. and Sander, C. (1993) Proc Natl Acad Sci U S A 90 (16): 7558-62; Rost, B. and Sander, C. (1994) Proteins, 19 (1) : 55- 72; PSIPRED program-Jones, D. T. (1999) J. Mol. Biol. 292 (2): 195-202), whereby the conformation of a protein or peptide molecule with respect to nearest-neighbour amino acids is analyzed. Also the frequency and type of amino acids surrounding the predicted indels and the indel complementarity region can be determined. Furthermore, the tertiary

and quaternary structure of the pathogen and host proteins can be compared to look for potential drug target sites on the pathogen protein that are not available on the corresponding host protein.

Molecular modeling of a protein structure can be performed using the SwissPDB Viewer package (Guex N. and Peitsch MC. (1997) Electrophoresis 18: 2714-2723) combined with the Molecular Operation Environment (MOE) software package (Version 2001. ou, Chemical Computing Group Inc. , Montreal, Canada) as described below.

Homology templates used for modeling can be retrieved from the Expasy server (www. expasy. ch) can provide the user with solved structures of the protein of interest or for similar proteins.

Example Characterization of Pathogenic EF-la Elongation factor-la (EF-loc) from Leishmania dofoiani is a Src-homology 2 domain containing protein tyrosine phosphatase-1 (SHP-1) binding and activating protein.

Persons skilled in the art would be able to prepare GST-SHP-1 construct, based on the expression and purification previously described by Jiao H. et al. ( (1996) Mol. Cell. Biol.

16: 6985-6992). cDNA encoding L. donovani EF-la was PCR amplified, cloned and sequenced. The deduced amino acid sequence shows that the leishmania protein is shorter than mammalian EF-la by a twelve amino acid deletion. In spite of this, the apparent molecular mass of leishmania EF-la is higher than its mammalian counterpart.

There is nearly complete sequence conservation amongst EF-la proteins from trypanosomatids, when compared to mammalian EF-la sequences. Consistent with this, the sub-cellular distributions of L. donovani EF-loc and host EF-lu, are strikingly different. In leishmania, the majority of EF-lot is Triton-X100-insoluble, whereas macrophage EF-la is predominantly cytosolic. Infection of macrophages with L. donova7zi causes redistribution of host as well as pathogen EF-la. In contrast to its predominant distribution in the Triton-insoluble fraction in promastigotes and amastigotes, leish7nania EF-loc was essentially completely cytosolic in infected macrophages. In addition, infection results in further redistribution of host EF-la to the cytosol. Protein modeling shows discrete 3-dimensional structural differences between

the proteins showing important structural differences between mammalian and trypanosomatid EF-1α and their role in the latter in pathogenesis.

Protein Function Analysis To examine whether leishmainia EF-1α associated with host SHP-1 in vivo, macrophages were infected with promastigotes for 14-16 hrs. Cytosolic fractions were then prepared from control and infected cells for immunoprecipitation of SHP-1.

Immune complexes were separated by SDS-PAGE under non-reducing conditions followed by transfer to nitrocellulose. Immunoblot analysis carried out using anti-EF-1α showed that leishmania EF-lol associated with SHP-1 in vivo, whereas the association of host-EF-1α with host SHP-1 was minimal. These findings show that leishrnania EF-loc is a selective SHP-1 binding protein in vivo during infection.

EF-1α was purified to near homogeneity from murine macrophages and from leishmania promastigotes. Binding assays performed using the purified proteins and GST-SHP-1 glutathione-sepharose beads showed that EF-lot *om leishmania bound directly and selectively to SHP-1 as comparatively little binding of host EF-1α was detected. In contrast, both leishmania and macrophage EF-1α bound directly and with similar affinities to calmodulin, a known EF-la binding protein. These findings show that the purified host EF-la was functionally intact and confirmed that leishmania EF-1 oc is a selective SHP-1 binding protein.

Purified EF-1α proteins from both macrophages and leishmania were examined for their ability to modulate SHP-1 phosphatase activity in vitro. Leishmania EF-loc, but not its mammalian homologue, was capable of activating GST-SHP-1 in vitro.

Purified EF-loc proteins introduced into macrophages using the protein delivery system Profect-1 (Targeting System, Santee, CA) resulted in SHP-1 activation in vivo. In contrast, activation of SHP-1 was not observed when purified macrophage EF-loc or BSA were used as control proteins. Delivery of purified, native leishn2cania EF-loc, but not the corresponding host protein into cells blocked induction of iNOS expression in response to cell treatment with interferon-y. Thus, leishmania EF-1α was able to recapitulate the deactivated phenotype of leishmania-infected macrophages. These results show that

leishnaania EF-la is a selective activator of SHP-1 capable of inducing macrophage deactivation.

EF-loc from vacuole bound leishmania must cross two membranes (the parasite plasma membrane and the parasitophorous vacuole) to access cytosolic SHP-1. We have been able to detect EF-1α as a tyrosine phosphorylated protein in promastigote growth medium in the absence of parasite lysis indicating that it is exported in some manner.

Furthermore, concentrated promastigote growth medium was able to activate SHP-1.

When leishmania infected macrophages are subjected to selective lysis to preserve phagosome integrity, followed by immunoprecipitation of cytosolic SHP-1, EF-la is identified in these complexes. Thus, these molecules do associate in vivo thereby providing an opportunity for activation of SHP-1. Our findings that the leishmania EF- lot is tyrosine phosphorylated and that the sequence contains two canonical ITIM motifs is consistent with the protein activity SHP-1 as a result of binding of tyrosine phosphorylated ITIM motifs to SH2 domains of SHP-1.

The Sub-cellular distribution of L. donovani EF-1α is distinct from macrophage EF-loc The distinct molecular mass of leishmania EF-loc as compared to the corresponding host protein suggested potential functional differences. To investigate this further, the sub-cellular distribution of EF-la in L. donovani promastigotes and macrophages was examined. Cytosolic, Triton-soluble and Triton-insoluble fractions were prepared from L. donovani promastigotes and macrophages, separated by SDS- PAGE, transferred to nitrocellulose and probed with anti-EF-loc. The results show that in leishnaajaia promastigotes, the majority of EF-loc was in the Triton-insoluble fraction, whereas in macrophages, EF-loc was predominantly cytosolic. The lower molecular mass band in the leishmaszia Triton-insoluble fraction (Ti) is most likely a proteolytic degradation product of EF-lot.

Sera from patients with visceral leishmaniasis recognize leishmania, but not host EF-lot. The results show that leishmania EF-loc is structurally distinct from human EF- la. EF-la was immunoprecipitated from Trition X-100 lysates of L. donovani promastigotes using anti-EF-la antibodies. Immune complexes were separated by SDS- PAGE under non-reduced conditions followed by transfer to nitrocellulose membranes.

Immunoblot analysis was carried out using either anti-EF-la, normal human sera or sera from visceral leishmaniasis patients. The results also indicate that leishmania EF-la was recognized by sera from patients infected with L. donovani. Importantly, EF-la immunoprecipitated from cells of the human mononuclear phagocytic cell line THP-1, was not recognized by sera from L. donovani infected patients. These findings show that the primary amino acid sequence differences between leishmania EF-l (x compared with human EF-1α are sufficiently distinct that corresponding structural differences contribute to the formation of epitopes in the leishmania protein that are processed and recognized by the immune system. It should also be noted that EF-1 oc of leishmania is a closely spaced doublet when separated on SDS-PAGE under non-reduced conditions.

Leishmania infection alters the subcellular distribution of both macrophage and leishmania EF-1 oc EF-l oc is known to be involved in a variety of cellular processes in addition to regulation of protein synthesis [Condeelis J. (1995) Trends Biochem. Sci. 20: 169-170 ; Kaur, K. J. and Ruben, L. (1994) J. Biol. Chez. 269: 23045-23050; Murray JW. et al.

(1996) J. Cell Biol. 135, 1309-1321 ; Ganatra JB. et al. (2000) Am. JOphthalmol.

129 : 166-172]. We examined the distribution of both host and leishmania EF-1 a during macrophage infection. For infection, macrophages were incubated with L. donovani stationary phase promastigotes for 16 to 18 h at a parasite-to-cell ratio of approximately 10: 1. This resulted in infection rates of> 95% with approximately four to eight promastigotes per cell. To determine the sub-cellular distribution of EF-la in non- infected and L. donovani infected macrophages, cytosolic, Triton-soluble and Triton- insoluble fractions were prepared as described in Experimental Procedures and analyzed for EF-1 oc by immunoblotting. In contrast to its predominant distribution in the Triton- insoluble fraction in promastigotes, leishnaania EF-Ioc was principally cytosolic in infected macrophages. Simultaneously, infection of macrophages with leishmania resulted in a redistribution of host EF-1 a to the cytosol, thereby increasing its overall abundance in this fraction. Similar results were obtained when cells were infected with leishmania amastigotes. The effect of leishmania infection on redistribution of EF-loc

was specific as phagocytosis by macrophages of Staphylococcus aureus did not cause redistribution of host EF-1α Protein Structure Analysis-Molecular Modeling of EF-loc Modeling of the structure of leishmania EF-lot was done using the SwissPDB Viewer package [Guex N. and Peitsch MC. (1997) Electrophoresis 18: 2714-2723]. A homology template search was done in which the sequence of leishmania EF-loc in Fasta format was downloaded into the software environment of SwissPDB. The EF-loc sequence was then submitted to the Expasy server (www. expasy. ch) to search against the database of proteins of known structure. The server returned the sequence of Saccharomyces cerevisiae EF-loc, (1G7CA) as a top homology search result. Three other PDB entries (1F60A, 1IJEA, 1IJFA) corresponding to EF-lot proteins from S. cerevisiae were also identified as possible modeling templates and were used for homology modeling of L. donovani EF-loc.

For homology modeling, the templates for 1G7C, 1F60A, 1IJEA, 1IJFA were downloaded from the ExPasy server and imported into SwissPDB program. The template backbones were superimposed by using"SwissPDB) Fit j Magic Fit"procedure. The structure of 1G7CA was used as a master template. To derive a preliminary structural estimate for EF-1α from L. donovani, the templates for 1G7C, 1F60A, 1IJEA, 1IJFA were superimposed and the sequence of the leishmania protein was fitted into the composite using"SwissPDB |Fit j Raw Sequence"procedure. Consecutive use of procedures"SwissPDB # Fit) Magic Fit","SwissPDB j Fit) Improve Fit"and"SwissPDB) Fit) Best"allowed the predicted structure to be improved. In order to refine this preliminary structure further, the side chains and terminal chains had to be modeled (SwissPDB usually truncates terminal and irregular regions). This was done using the Molecular Operation Environment (MOE) software package (Version 2001.01, Chemical Computing Group, Inc., Montreal, Canada). The homology model of leishmania EF- la (previously built by the SwissPDB-Viewer) was used as a template and the curated PDB file 1G7C from the PDB database provided with the MOE package was used as a second template. The 1G7C template from the curated PDB database was imported by command"MOE # File # Protein Database"and the sequence of leishmania EF-lorv and its

preliminary SwissPDB homology model were downloaded using option"MOE j File j Open". The MOE Sequence Editor was then used for iterative sequence alignment with Gonnet substitution matrix. Secondary structure elements were not used for the sequence alignment. The MOE command"Seq. Editor) Homology # Align"was used to perform the alignment. Homology modeling of leishmania EF-lot was then carried out with the highest degree of protein structure optimization. The previously derived SwissPDB model of the leishma7lia protein was used as a template. The final structure was derived as a Cartesian average of the top ten scored, non-minimized intermediate models. The estimated top ten homology models built for leishmania EF-la were saved in MOE database *. mdb format to be viewed by the MOE database Viewer called by"MOE | Open | *. mdb". The final 3-D structure of the protein was averaged over the top ten scoring models.

Once the refined 3-D model of the EF-1α protein from L. donovani was created, the homology modeling procedure was repeated on a modeling template of EF-1α from S. cerevisiae (PDB file 1 G7CA). The modeling routine was carried out in the exact same way as previously described for L. donovani's EF-lot with all numerical parameters the same. This step was taken in order to justify the accuracy of the homology model of leishmania EF-1α and to build its terminal chains not previously modeled by the SwissPDB Viewer. The homology model of the protein EF-la from L. donovani created on the 1G7CA template by MOE resembled the SwissPDB model with very high accuracy. The two structures of the leishmania protein created by MOE on the templates of 1G7CA and of the SwissPDB model had a high degree of resemblance with a calculated RMSD between the two structures of 0.69A. Based upon this analysis, the 3- D structure of EF-1α from L. donovani built using MOE on a template of the SwissPDB model was accepted as final. Lastly, since EF-1α proteins from Mus musculus and Homo sapiens share 99.8% identity and have 81.1% sequence identity with 1G7C, homology modeling of these proteins was done in the same manner as for L. donovani EF-lot.

Cloning of L. donovani EF-1 oc cDNA and determination of the deduced amino acid sequence

Alignment of EF-1α genes from L. braziliensis, T. cruzi, T. brucei and others revealed a high degree of conservation. PCR primers were designed based upon the sequence of the L. braziliensis gene and used to amplify the coding sequence of the L. donovani EF-1 oc gene from cDNA of L. donovani promastigotes. The predicted amino acid sequence of EF-1 oc of L. donovani was compared with other trypanosomatids (T. brucei, T. cruzi) and with the human sequence. The multiple alignment comparison showed that a high degree of homology extended through the complete sequence. The GTP-binding consensus motifs and the three lysine residues that are usually post- tranlationally modified were found to be conserved in L. donovani. Notably however, there were several important differences observed in the trypanosomatid sequences as compared to the human sequence. For example, it was remarkable to note at position 151-152 the replacement of glycine and valine in the human sequence with cysteines in the trypanosomatid sequences, suggesting possible differences in folding of the proteins.

In addition, human EF-1α was found to contain twelve extra amino acids when compared with trypanosomatid EF-1 oc sequences. The protein sequence of L. donovani shows the presence of two previously unrecognized putative immunoreceptor tyrosine-based inhibitory motifs (ITIMs). These specialized motifs are known to be present in many signaling molecules with the capability of binding to Src homology 2 domains (SH2 domains) [Berg KL. et al. (1998) Oncogene 17: 2535-2541; Frearson JA. and Alexander DR. (1997) BioEssays 19: 417-427; Blery M. et al. (2000) Hum. Immunol. 61: 51-64].

The molecular mass of trypanosomatid EF-1α was found to be higher than mammalian EF-loc. Initial immunoblot analysis of EF-la in Triton X-100 cell lysates of L. donovani promastigotes and the human carcinoma cell line A431 showed that the apparent molecular size of EF-loc of L. donovani was distinctly higher than the human homolog. To determine whether this was a specific characteristic of the leishmania protein or a more general property of the trypanosomatid family of which L. donovani is a member, Triton X-100 cell lysates of several trypanosomatid and mammalian cell lines were separated on SDS-PAGE, transferred to nitrocellulose and probed with anti-EF-loc antibodies. The results show that the apparent molecular mass of EF-loc from trypanosomatids is higher than mammalian EF-loc. The apparent molecular size of EF- loc from trypanosomatids was found to be approximately 56 kDa as compared to 50 kDa

for EF-1α of mammalian origin. The molecular mass of EF-loc did not change when Triton X-100 lysates of L. donovani promastigotes at different stages of growth (exponential and stationary phases) were analyzed by immunoblotting.

Analysis of Molecular modeling of EF-la To model and analyze the 3-D structure of leishmania EF-lot, its amino acid sequence was scanned against the database of the proteins with known 3D structures.

This analysis showed that L. donovani EF-othas 75% identity with EF-la from S. cerevisiae. The 3-D structure of S. cerevisiae EF-loc was previously determined experimentally (PDB entry 1G7C) [Andersen GR. et al. (2001) Nat. Struct. Biol. 8: 531- 534) ] and was used as a template for L. donovani EF-1α protein modeling. The estimated degree of identity (above 75%) allowed for accurate modeling of L. donovani EF-loc based upon the 1 G7C template. The 3-D structure of the main body of L. donovani EF- 1 a was constructed by homology modeling and side chains were generated by averaging a number of computed models. Protein sequence alignment also showed that EF-1α from S. cerevisiae shares 77.3% sequence identity with EF-la from Mus musculus and 77.3% with EF-1α from Homo sapiens. The 3-D structures for these proteins were also modeled and further refined based upon the template of 1G7C. Comparison of the models shows that the 3-D structures of EF-loc proteins from S. cerevisiae, L. donovani, Mus musculu and from Homo sapiens resemble each other closely. Without taking into account the differences in irregular C-terminal loop regions, the only notable difference in the structures of the human and leishmania proteins is attributed to an isolated hairpin fragment corresponding to the insertion of twelve amino acids in the H. sapien's protein sequence at position 214, that is missing from L. donovani EF-lot. The structure of this twelve amino acid fragment was identified as a hairpin motif comprised of two anti- parallel ß strands each of which is four amino acids in length. L. donovani EF-la has a proline residue at position 214 corresponding to point of insertion of the hairpin fold in the H. sapiens EF-loc. This proline likely functions to stabilize the backbone of L. donovani EF-1α by preserving local structural similarity with EF-lafrom H. sapiens.

However, this proline"stitching"at the point of the deletion in the 3-D structure of L. donovani's EF-loc is likely not able to compensate fully for the nearly 5.4A wide gap

corresponding to the distance between top edges of the hairpin, and some distortions in flexible, irregular chains surrounding the deletion site in the leishmania protein may occur.

The cloning and characterization of EF-la from L. donovani have now provided novel information to show clearly that the leishmania protein has unique properties that distinguish it from macrophage EF-la. The deduced 449 amino acid sequence for the leishinania protein was observed to share identity of >90% with other known EF-la proteins from trypanosomatids, whereas identity was found to be 75.5% with the human protein. Despite this high degree of overall protein sequence homology, important amino acid substitutions such as replacement of glycine and valine at positions 151-152 of the human sequence by cysteines in the leishinania sequence were found. In addition, a twelve amino acid deletion in the primary sequence of the L. donovani sequence was identified. Protein modeling predicts significant differences in 3-D structure when comparing the leishmania protein with the human homologue. Leishmania and mammalian EF-loc's were also observed to have distinct apparent molecular sizes and subcellular distributions. The results clearly show that the distinct molecular mass of leishmania EF-la is a general property shared by homologues obtained from other trypanosomatids. Furthermore, the leishmania protein, but not its mammalian homologue binds to and activates SHP-1 These distinct properties are influenced by the differences in 3-D structure.

EF-1 oc is known to contain a large number of post-translational modifications, whose roles have not been established. Most notably, EF-l a is modified by methylation of lysine residues [Damen JE. et al. (1995) JBiol. Chez. 270: 23402-23408] and glycerolphosphoryl-ethanolamine additions [Stein RC. and Waterfield MD. (2000) Mol.

Med. Today 6: 347-357; Hoedemaeker FJ. et al. (1999) JMol. Biol. 292: 763-770; Yohannan J. et al. (1999) JBiol. Chem. 274: 18769-18776] and the appropriate residues are present in the L. donovani protein. Sequence analysis ofL. donovani EF-loc showed that the protein has two putative ITIM motifs (xxYAxV) that are conserved and analysis with 4G10 antibodies indicated that it is tyrosine phosphorylated. These specialized motifs are known to be present in many signaling molecules and have the potential to bind to SH2 domains [Berg KL. et al. (1998) Oncogene 17: 2535-2541; Frearson JA. and

Alexander DR. (1997) BioEssays 19: 417-427 ; Blery M. et al. (2000) Hum. Immunol.

61 : 51-64]. It has been shown that a phospholipase C-yl construct containing both SH2 and SH3 domains directly binds to EF-1α [Vanhaesebroeck B. et al. (1999) Nat. Cell <BR> <BR> <BR> <BR> Biol. 1:69-71]. These motifs account for the binding of leishmania EF-1α to SHP-1,<BR> <BR> <BR> <BR> <BR> <BR> whereas overall differences in 3-D structure accost for why the mammalian homologue does not do so, in spite of the fact that it also contains the same ITIM motifs. The presence of antibodies against leishmania EF-1α (but not the human homologue) in sera of patients with visceral leishmaniasis support further the conclusion that leishmania EF- 1 a is structurally distinct from host EF-1α.

Tlzree-dimensional structure of EF-1α from L. donovani was modeled and compared with EF-la of H. sapiens. In spite of the overall high degree of primary amino acid sequence similarity, a hairpin of twelve amino acids was modeled that was unique to the H. sapiens protein. To investigate functional signiiica-nce of the indel (hairpin from the human form), its sequence -GWKVT217RKDGNASGT- (available from genBank) was searched against the PROSITE database that contains information about protein motiis with defined functions. This analysis showed that S. sapien's EF-la contains potential consensus phosphorylation sites (PROSITE matches PS00005 and PSt0006) Por protein kinase C (Kishimoto A. et al. (1985) JBiol. Chen. 26012492-12499) and casein kinase 11 (Pinna, LA. (1990) Blochim. Biophys. Acta 1054: 267-284) at threonine217 in the hairpin loop. These findings show that the hairpin loop present in EF-1α proteins from H. sapiens and other higher eukaryotes and missing in L. donovani EF-1α, contributes to the experimentally observed differences in their : function and subcellular distribution.

The crystal structure of EF-1α proteins from S. cerevisiae reveals the close proximity of the 20A long hairpin to the main body of me protein. Thus, the side chains of the hairpin form a complex network of hydrophobic and polar interactions involving me a-helix formed by ammo acids 226-231, the ß-strand formed by residues 233-236 and the low complexity region of amino acids 182-191 in the main body of the protein. For example, Glu215 in the hairpin reaches a proximity of 2A with the side chain of Letll84 in the main part of the protein implying the formation of hydrogen bonds between the residues. The homologue also shows these structural properties.

The remarkable absence of these twelve residues and the hairpin formed by them in the structure of leishmania EF-loc, provides the basis to design small molecule inhibitors that bind specifically to the regions of the leishmania protein otherwise shielded by the hairpin fragment in the H. sapiens protein. Specifically, the putatively "unshielded"region of leishmania EF-loc contains several highly polar residues such as Asp"', Glu224, Met222, Lysl87 and Arg 189 with their charged side chains pointed directly toward the predicted location of the missing hairpin. One or more of these residues are sites for drug attack. Conversely, the bulky amino acids, Trp21°, Phe2n, Trp2l4 and Tyr2l7 in the hairpin contribute to blocking an attacking reagent from reaching the corresponding structural region of the Homo sapien's homologue.

EF-loc may be selectively inactivated in leishma71ia, without interfering with the host EF-1α homolog, considering the essential requirement for EF-la for survival and its role as a virulence factor. The absence of the hairpin structure from EF-loc sequences of other pathogenic trypanosomatids (e. g. T. cruzi, T. brucei brucei and T. congolense) and unrelated protozoa such as Entameoba histolytica and the other protozoans referred to herein, shows that such drugs will have affect on an important range of widely prevalent pathogens.

The results show that leishmania infection causes the redistribution of both host and pathogen EF-loc. During infection, leishmania EF-loG became principally cytosolic in nature in contrast to its predominant Triton-insoluble distribution in extracellular promastigotes. Simultaneously, infection of macrophages with leishmania resulted in the redistribution of host EF-1α to the cytosol, thereby increasing its overall abundance in this cellular compartment. The results show that leishmania EF-locis principally cytosolic in infected cells. Similar results were obtained when cells were infected with either promastigotes or amastigotes, indicating that redistribution of leishmania EF-loeis not due to parasite transformation during infection. Redistribution of host and pathogen EF-loc during leishmania infection appears to be specific for L. donovani infection, since redistribution of host EF-1α was not observed when cells had ingested fixed S. aureus.

C) Targeted Drug Development

Once the structure of the pathogen protein and particularly the pathogen specific region is characterized as described above. Bioinformatic programs may be used to screen the conformations of known small molecules for potential fit within the indel complementarity region or the indel itself depending on whether the indel occurs on the host (then target indel complementarity region) or on the pathogen (then target the indel region itself). Furthermore, potential binding moieties may be incubated in solution with the target protein crystals, and the complementarity region or the indel region may be once more defined by crystallography to determine fit of the therapeutic or drug molecule. If the structure of a target pathogen protein is determined by molecular modeling or by X-ray crystallography, then a person skilled in the art would be able to then design a three-dimensional shape of a putative binding moiety (e. g. see: U. S. Patents 6,127, 524 and 6,168, 928. ).

Example-leishmania EF-la The"unshielded"region of pathogenic EF-la, which is concealed by a hairpin loop in the human protein, contains several highly polar residues with exposed side chains. Small molecules and peptides may be designed (using conventional modeling techniques and available software) to specifically bind to this unshielded region. Such a specific binding moiety will have a binding site containing one or more binding elements, which may be one or more chemical groups capable of forming a hydrogen or ionic bond or participate in a hydrophobic interaction with one or more contact residues in the unshielded region of EF-la. Preferably, two or more such bonds and/or interactions will exist. The aforementioned bonds may exist between a group on the specific binding moiety and atoms in the polypeptide backbone or in an amino acid side chain at the unshielded region, in the presence of or the absence of water molecules between the binding moiety and the unshielded region. Preferably, the distance between a binding element in the moiety and a contact residue will be 5 Angstroms or less, more preferably from 1-4 or from 2-3 Angstroms. Preferably, a contact residue will be selected from one or more amino acids in the unshielded region, capable of forming a hydrogen bond or which have polar side chains, preferably amino acids corresponding to one or more of Asp 218 or 221, Glu 186 or 224, Met222, Lys 187 and Argl89 of leis » 7nania EF-la. The specific binding moiety may have the capacity of binding other regions of EF-la by hydrophobic

or polar interactions, particularly with the regions corresponding to amino acids 182-191 and 226-236 of yeast EF-la (such as Leu 184) or amino acids 158-168, 186-194, and 215- 224 of leisAmania EF-la. The atomic coordinates of EF-la as determined by molecular modeling or by X-ray crystallography may be used to then design a three-dimensional shape of a putative binding moiety (e. g. see: U. S. Patents 6,127, 524 and 6,168, 928).

Characterization of EF-la as described above shows that antibodies specific for the pathogenic variant may be generated. In addition to small molecules/peptides, specific binding moieties according to this invention may be antibodies or antibody fragments specific for the unshielded region of pathogen EF-la or which otherwise specifically bind to the pathogen form. Antibodies and antibody fragments may be generated by well known techniques.

The specific binding moieties described above may cause a change in function of pathogenic EF-la (e. g. by affecting SHP-1 binding or by affecting EF-la's role in protein translation) or the moiety may simply bind with specificity to the pathogenic form. In the latter instance, the moiety may be used as a targeting ligand in assays for the label or pathogenic form (e. g. with a radiolabel or other moiety which facilitates detection) or as a targeting ligand for therapeutic moieties that themselves bring about an affect on the pathogenic protein.

Ahti-pathogenic EF 1 a Antibodies. Antibodies towards the three-dimensional surface differences of the leishma7lia EF-lot protein will have efficacy since they may interfere with either one or both of two important functions that enable the pathogen to infect humans. The first function is the fundamental role of EF-la in pathogen protein translation. The second is the role of EF-la as a virulence factor.

The three-dimensional model of the leishfnafzia identifies the peptide sequences that are normally concealed (indel complementarity region) by the hairpin loop (indel) present on human EF-loc. Using publicly available protein structure visualization programs, short peptide sequences derived from leishma7zEa EF-la may be selected for development of an antibody using standard techniques for monoclonal antibody (mAb) production. Three sequences are: 1. amino acids 215-224 TLLDALDMLE 2. amino acids 186-194 EKVRFIPIS 3. amino acids 158-168 KTVTYAQSRYD

Specific mAbs may be used to investigate the importance of the indel complementarity region on leishmania EF-loc using independent assay systems which determine: (1) whether they interfere with the function of leishma7zia EF-loc in in vitro translation assays, (2) whether they block the interaction of leishmania EF-loc with SHP- 1 in vitro, and (3) whether upon introduction into leishmania infected macrophages they affect parasite viability.

Recombinant peptideslproteins to target EF-1 a for proteolysis. Given the differential epitopes displayed on the otherwise highly conserved EF-lot proteins of leishmania and humans, this strategy involves the construction of a moiety that binds and neutralizes the exposed epitope of leishmania EF-l (x. In the case of a moiety that binds but does not inhibit the protein synthesis function or the virulence factor properties of EF- la, an alternative strategy utilizing macrophage intracellular trafficking may be used to destroy the pathogen EF-la protein.

Examples of two strategies are ones which utilize aspects of the ubiquitin pathway responsible for the majority of protein turnover within a eukaryotic cell. Typically ubiquitin is ligated to proteins targeted for destruction and serves as a marker for transport to lysosomes and subsequent proteolysis. Both strategies take advantage of the capacity of a synthetic indel peptide based upon the human EF-loc sequence to selectively bind the exposed epitopes available on the surface of leishnmania EF-loc, but not on the human protein. The synthetic indel will contain either a terminal signal for ubiquitination, or the protein sequence for ubiquitin.

Chimeric EF-1 a indel One of the ubiquitination complexes that has been well characterized governs the turnover of ItBoc, the inhibitor of NFKB. NFKB controls expression of many genes associated with inflammation. The protein IKBoc is recruited to the ubiquitination complex by virtue of a 10 amino acid sequence (hereafter referred to as'recruitment peptide') that is phosphorylated in response to inflammatory signals [Sakamoto KM. et al.

(2001) PNAS 98 (15): 8554-9]. Thus, at the onset of inflammation upon infection, the inhibitor to inflammatory gene expression is destroyed, and upregulation of NF-KB controlled genes expression occurs and the recruitment peptide is activated. The

recruitment peptide may be added onto the C-terminal of a specific binding moiety as described above which may be the actual 12 amino acid indel, joined by an amino acid to the recruitment peptide. The chimeric indel may be delivered to leishmania infected macrophages using either liposomes or a protein targeting reagent such as Profect-1TM.

LeisZl7nania viability in macrophages may be monitored and compared to control treated cells (introduction of an irrelevant peptide sequence of identical size and linked to the recruitment peptide).

A second strategy uses expression of a recombinant peptide comprised of an N- terminal ubiquitin sequence and a C-terminal indel peptide. The C-terminal sequence of ubiquitin may be mutated from a glycine to an alanine to prevent cleavage of the ubiquitin protein [Tellam J. et al. (2001) Journal of Biological Chemistry 276 (36): 33353- 60]. The cDNA for human ubiquitin may be amplified by PCR from a human cDNA library using specific 5'and 3'oligonucleotide primers to facilitate cloning of the ubiquitin gene into the prokaryotic expression vector pBlueBac4. 5/V5-His Transfer Vector (Invitrogen), permitting expression in an insect cell line. The 3'oligonucleotide will contain the appropriate mutation to convert the glycine76 codon to alanine. The nucleic acid coding sequence for the indel may be synthesized as two complementary oligonucleotides and ligated in frame with the ubiquitin gene to allow for expression of a recombinant ubiquitin-indel protein. Purified ubiquitin-indel protein may be delivered to leishmania infected macrophages as described above for chimeric proteins, followed by assessment of parasite survival.

X-ray crystallography Persons skilled in the art would know how to express pure EF-loc in a recombinant system such as baculovirus and how to purified by affinity chromatography.

A full length L. donovani EF-la gene may be PCR amplified using oligonucleotide primers that contain restriction endonuclease sites to facilitate ligation into a baculovirus expression vector. A baculovirus system is preferable over prokaryotic expression systems [Button, LL. et al. (1993) Gene 134 (1) : 75-81], because the EF-Ia protein appears to undergo post-translational modifications, including methylation of lysine residues and glycerolphosphoryl-ethanolamine additions, not possible in bacteria. Human EF-la may be cloned, expressed, and purified using this methodology.

The L. donovani EF-la-EK-His coding sequence may be similarly sub-cloned into an expression vector such as pBlueBac4.5 (Invitrogen) and cotransfected with the Bac-N- Blue vector (Invitrogen) into Sf9 insect cells and the viral supernatant harvested at day three. Plaque assays may be performed to isolate pure virus that will be screened by PCR to identify recombinant virus. High titer virus stocks can be prepared and used to infect Sf9 insect cells for high level expression of L. donovani EF-la [Button, LL. et al. (1993) Gene 134 (1) : 75-81]. Cell lysates may be prepared and the rEF-la purified by affinity chromatography using ProBond columns (Invitrogen). After elution, rEF-la may be treated with enterokinase (enteropeptidase) (New England Biolabs) to remove the His tag, purified by FPLC gel exclusion chromatography and used for functional analysis.

Affinity purified, baculovirus-expressed EF-la may be incubated with GST-SHP-1 to examine its ability to bind to and activate the enzyme by phosphatase assay as described above. Synthetic phosphopeptides derived from the primary sequence of the two ITIMs of EF-la, including five amino acids on either side of each, may be biotin conjugated and used to bind to and activate SHP-1 in vitro. These peptides may be compared with peptides in which tyrosine has been replaced by phenylalanine. The peptides may also be employed to compete with full length EF-la for binding to SHP-1. PCR-based, site- directed mutagenesis as done previously [Button, LL. et al. (1993) Gene 134 (1) : 75-81] may be used to produce full length protein in which either one or both tyrosines have been changed to phenylalanine to abolish SHP-1 binding.

Crystals of leishmania and human EF-lot may be grown following known protocols for yeast EF-la crystallization. Crystals of leishmania and human EF-la may be grown by the hanging drop method, for example by combining equal volumes (2ut) of the protein stock and mother liquor (e. g. 20% polyethylene glycol 8000,0. 1 M cacodylate, pH 6.5) at room temperature. Crystals may be subjected to vitrification in 30% polyethylene glycol 8000,0. 1 M cacodylate, pH 6.5 and 15% glycerol using liquid propane. Diffraction data may be collected, for example by using a Mar345 Image Plate detector with osmic mirrors mounted on a Rigaku RU-200 rotating anode x-ray generator.

Data may be processed using HKL programs. Molecular replacement may be performed using CCP4 software, and subsequent refinement of the model may be done using the CNS program. An alternative expression system for the synthesis of recombinant proteins in leishmania may also be used [Joshi PB. et al. (1995) Gene 156 (1) : 145-9;

Laban A. and DF. Wirth (1989) PNAS 86 (23): 9119-23; LeBowitz JH. et al. (1990) PNAS 87 (24): 9736-40; Kelly JM. et al. (1992) Nucleic Acids Research 20 (15): 3963-9], followed by purification and x-ray crystallography.

Bioinformatic programs may be used to screen the conformations of known small molecules for potential fit within the indel complementarity region of leishmania EF-l ot [Ooms F. (200) Current Medicinal Chemistry 7 (2): 141-58; Kurogi, Y. and O. F. Guner (2001) Current Medicinal Chemistry 8 (9): 1035-55; Gradler U. et al. (2001) Journal of Molecular Biology 306 (3): 455-67; Aronov AM. et al. (2000) Biochemistry 39 (16): 4684- 91]. Potential binding moieties may be incubated in solution with EF-la protein crystals, and the complementarity region once more defined by crystallography to determine fit of the therapeutic molecule.

Although the foregoing invention has been described in some detail by way of illustration and example for purposes of clarity of understanding, it will be readily apparent to those of skill in the art in light of the teachings of this invention that changes and modification may be made thereto without departing from the spirit or scope of the appended claims. All patents, patent applications and publications referred to herein are hereby incorporated by reference.

Table 1. Pair wise percentage residue identity between sequences of lEF-eprotein from 4 organisms.

EFIa EF11 HU1G7C. A AAH0406 EF-la (Leishmania donovani) 100.0 75.5 75.0 75.5 EFl l_HUMAN (Homo sapiens) 77. 7 100.0 81.1 99.8 1G7C. A (Sacc7laromyces Cerevisiae) 73. 5 77.3 100.0 77.3 AAH0406 (Mus musculus) 77. 7 99.8 81.1 100.0