CROSS REFERENCE TO RELATED APPLICATIONS

[0001] This application claims priority to U.S. Application Serial No. 62/166,565, filed May 26, 2015, the content of which is herein incorporated by reference in its entirety.

STATEMENT OF GOVERNMENTAL INTEREST

[0002] This invention was made with Government support under Contract No.

K01EB016071 awarded by the National Institutes of Health. The Government has certain rights in the invention.

BACKGROUND

[0003] Single cell analysis is important for understanding how cells respond to drugs and other perturbations because phenotypic responses are asynchronous and cells are often both genetically and epigenetically heterogeneous. In the absence of artificial or external perturbations, the natural processes of cellular differentiation during

development and malignant transformation in cancer also occur asynchronously and give rise to phenotypic heterogeneity. While single cell RNA-Seq is emerging as a high- dimensional and increasingly scalable tool for assessing phenotypic heterogeneity, many aspects of cellular phenotype cannot be inferred from the transcriptome alone. However, high-content imaging by microscopy could greatly expand the space of observables in a single cell analysis, particularly if it can be integrated with single cell RNA-Seq. High- content imaging assays using optical microscopy provide access to numerous phenotypic observables for cellular metabolism, protein localization, protein synthesis, cell cycle states, and cell signaling all with single cell resolution. However, existing tools for merging single cell imaging and sequencing are expensive, low-throughput, and incompatible with short-term cell culture and stimulation. The instant disclosure addresses this issue by combining a highly scalable microfluidic platform for single cell RNA-Seq and imaging with associated methodologies that allow for the association of sequence barcodes for inexpensive pooled library preparation with optical barcodes in devices described herein.

SUMMARY

[0004] The disclosure allows for devices, systems, and methods relating to genome- wide profiling of RNA from hundreds to thousands of individual cells in parallel for only a few cents per cell. The devices, systems, and methods described herein address the scalability problem of parallel preparation of low-input single cell libraries for RNA sequencing. Devices, systems, and methods described herein allow for parallel RNA profiling of individual cells in a device that is compatible with short term cell culture, drug stimulation experiments, and high-content fluorescence imaging. In an aspect, this is because the cells are physically segregated into microwells but still in sufficiently close proximity to communicate via diffusible factors.

[0005] In an aspect, the disclosure provides for a device or system comprising

(a) one or more mRNA capture beads;

(b) one or more cell-identifying optical barcodes;

(c) a plurality of chambers, microchambers, or microwells comprising one or more mRNA capture beads and/or one or more cell-identifying optical barcodes; and

(d) wherein the plurality of chambers, microchambers, or microwells comprising one or more mRNA capture beads and/or one or more cell-identifying optical barcodes are configured for reversible sealing.

[0006] In an aspect, the disclosure provides for a method of drug discovery, drug profiling, and/or drug testing comprising

(a) combining one or more mRNA capture beads with one or more cell- identifying optical barcodes;

(b) adding one or more mRNA capture beads with one or more cell-identifying optical barcodes to a plurality of chambers, microchambers, or microwells;

(c) wherein the plurality of chambers, microchambers, or microwells including one or more mRNA capture beads and one or more cell-identifying barcodes are configured for reversible sealing; and (d) adding one or more drugs to the plurality of chambers, microchambers, or microwells configured for reversible sealing; and

(e) adding one or more cells to the plurality of chambers, microchambers, or microwells that are configured for reversible sealing.

[0007] In an aspect, the one or more optical barcodes include one or more optical barcode sequences. In another aspect, the one or more optical barcodes include one or more fluorescent dyes including different colors and/or intensities.

[0008] The disclosure further provides for method of drug discovery, drug profiling, and/or drug testing described herein comprising (f) adding a buffer to the plurality of chambers, microchambers, or microwells configured for reversible sealing. In an aspect, the buffer includes a lysis buffer.

[0009] The disclosure further provides for a method of drug discovery, drug profiling, and/or drug testing comprising analyzing said cell-identifying optical barcode sequences by fluorescence.

[0010] In another aspect, the device or system of described herein comprise a plurality of chambers, microchambers, or microwells attached to a polysiloxane substrate. In yet another aspect, said polysiloxane substrate comprises polydimethylsiloxane.

[0011] In another aspect, the device or system described herein comprise cell- identifying optical barcode sequences are adapted for florescence detection.

[0012] In yet another aspect, the device, system, or methods comprise cells and said cells are not arranged in a droplet configuration.

[0013] In another aspect, the disclosure provides for a method for single cell RNA capture and sequencing, wherein said method comprises

(a) introducing at least one cell into a microwell, wherein the microwell is

attached to a first substrate that faces a second substrate and wherein oligo primers are attached to the surface of said second substrate;

(b) hybridizing mRNA molecules to the surface of said second substrate;

(c) adding at least one buffer to the microwell; and

(d) contacting said first substrate with said second substrate, thereby creating a seal between said first substrate and said second substrate. [0014] In yet another aspect, the disclosure provides for a device for single cell

RNA capture and sequencing, wherein said device comprises a plurality of microwells; wherein said plurality of microwells are attached to a first polysiloxane substrate; a second substrate comprising glass that faces said first substrate, wherein oligo primers are grafted onto to the glass surface of said second substrate.

[0015] The disclosure further provides for a microwell comprising a composition comprising one or more mRNA capture beads; one or more cell-identifying optical barcode sequences; one or more cells, one of more buffers.

BRIEF DESCRIPTION OF THE FIGURES

[0016] FIG. 1A provides for florescence imaging of a microwell array loaded with optically barcoded beads after introduction and hybridization of a single optical barcode probe. The brighter beads contain optical barcode oligonucleotides that are complimentary to the probe. FIG. IB provides for a series of 12 fluorescence images from each of two beads in the microwell array device after introduction of each of 12 barcoded probes. When a probe is complimentary to optical barcode oligonucleotides on the bead surface, the bead becomes fluorescent, resulting in the 12-digit binary code that corresponds to a sequence attached to the capture oligonucleotides on the bead.

[0017] FIG. 2A shows schematic and fluorescence imaging data for single cell

RNA capture on beads. (FIG. 2A) For mRNA capture on polymer beads, the microwell array is fabricated in a thin PDMS layer on top of a glass slide or coverslip with a microfluidic flow channel above. Cells are first deposited in the microwell array by gravity followed by beads (while circles) cov-2Balently functionalized with oligo(dT) primers (orange circular outlines). A lysis buffer is introduced followed by rapid displacement of fluid in the channel with oil, which conformally seals the array. Single cell ly sates (green) become trapped in individual microwells and mRNA hybridizes to the oligo(dT) on the beads (red circular outlines). FIG. 2B are close-up images of single cell RNA capture on beads. The top panel is a bright field/fluorescence overlay of a microwell array in which four microwells contain a bead, but only one contains both a bead and a cell (fluorescently labeled with live stain). The middle panel is a fluorescence image of the array after RNA capture, reverse transcription, and staining with SYTOX Orange. Note that the bead associated with a cell is significantly brighter than the other beads. The bottom panel is a fluorescence image of beads in an array from a negative control experiment involving no RNA or cells, showing that the beads have a certain level of background fluorescence in the presence of stain, which explains the majority of the background signal observed in the beads with no cell in the middle panel.

[0018] FIGS. 3A and 3B provide for combinatorial scheme for synthesis of barcoded capture beads. As illustrated in FIG. 3A, beads are first attached to a set of barcoded oligonucleotides in a multi-well plate, pooled into a single tube, and then redistributed into a second multi-well plate for combinatorial addition of a second barcode sequence and capture site (oligo(dT)). FIG. 3B provides for detailed molecular biology for solid-phase, combinatorial barcode synthesis. A first barcode sequence is copied onto a dual-biotinylated oligonucleotide containing the T7 promoter sequence and a partial Illumina adapter using DNA polymerase. The resulting double-stranded DNA is conjugated to streptavidin-coated beads, and the non-biotinylated strand is removed. After pooling and expanding the beads, a second reaction is used to add a second barcode sequence and oligo(dT) by priming off of a universal anchor sequence that follows the first barcode.

[0019] FIGS. 4A - 4C provide for a schematic and fluorescence imaging data for single cell RNA printing. As shown in FIG. 4A, cells are first deposited in the microwell array by gravity. The glass surface opposite the microwell array is covalently

functionalized with oligo(dT) primers for mRNA capture (orange line). The device is then rapidly and conformally sealed against a glass surface in the presence of lysis buffer, flipped over, and held in a sealed position using negative pressure. Single cell lysates (green) become trapped in the sealed microwells, and mRNA hybridizes to the oligo(dT) primers on the glass surface, resulting in single cell mRNA "prints" (red lines). As shown by FIG. 4B an array of single cell mRNA prints on a glass coverslip generated using the device in FIG. 4A and imaged after on-chip reverse transcription. The double-stranded RNA/DNA hybrids are stained with SYTOX Orange, an intercalator dye and imaged on the glass surface. >96% of the prints result from individual cells. Note that the bright spots in the image that are not registered with the array originate from genomic DNA aggregates that were not fully removed by DNase digestion. FIG. 4C shows close-up images of single cell RNA printing. The left-most panel is a bright field image of three cells in individual microwells of the array, the middle panel is a fluorescence image of the corresponding RNA prints on the glass surface after reverse transcription and staining with SYTOX Orange, and the right-most panel is a fluorescence image of the glass surface after RNase digestion, demonstrating that the fluorescent prints originate from captured RNA.

[0020] FIGS. 5A-5C provide for an example flow cell device for single cell

RNA-Seq. (A) Graphical representation of our five-lane microwell array flow cell device for single cell RNA-Seq. FIG. 5B shows a schematic of on-chip steps for single cell RNA-Seq. After depositing cells, barcoded capture beads (barcode sequences represented as different colors), and sealing as in FIG. 2A, single cell lysates (green) are trapped in individual microwells and mRNA hybridizes to the barcoded capture beads. The device is unsealed and rapidly washed by flow before on-chip, solid-phase reverse transcription and second-strand synthesis followed by elution and pre-amplification of the pooled library by in vitro transcription. FIG. 5C shows a montage of fluorescence images from part of one lane of the device in FIG.5A showing beads (red) and cells (blue) loaded in the array. Note that this image was acquired following cell lysis while the device is sealed, and so the blue live stain fills the entire volume of the corresponding microwell and is confined to the microwell by sealing.

[0021] FIGS. 6A and 6B show an analysis of single cell RNA-Seq data. FIG. 6A shows a gene body distribution for uniquely mapped reads showing that we are primarily sequencing the 3 '-end of transcripts, as expected. FIG. 6B shows a histogram of the number of genes detected per cell for the 396 single cell profiles used in all subsequent analysis of Experiment 1 and 247 single cell profiles used in all subsequent analysis of Experiment 2.

[0022] FIGS. 7A - 7C provide for (FIG. 7A) a comparison of single cell median and population- level RNA-Seq profiles for cells originating from the U87-exclusive lane in Experiment 1. Each data point was obtained by constructing a median profile from a given number of cells and repeating this ten times by random sampling with replacement to obtain a median Pearson correlation coefficient and error bar (SEM). This exercise was repeated for comparison to both the U87 and MCFlOa bulk RNA-Seq profiles to demonstrate better concordance between the U87 single cell profiles and the U87 bulk profile. FIG. 7B provides for the same as FIG. 7A, but for single cell profiles in the MCFlOa-exlusive lane. As shown in FIG 7C, differential expression analysis was conducted to obtain cell type-specific gene sets for the U87 and MCFlOa cells based on single cell profiles from the pure-cell lanes. Here, we show a histogram of log-ratio of the coefficients of variation (CVs) for the cell type-specific gene sets between the mixed lane profiles and the profiles from the respective pure lanes. As expected, the heterogeneity given by CV is greater for cells in the mixed lanes than in the cell type- exclusive lanes for the cell type-specific genes.

[0023] FIGS. 8A - 8C provide for cell type separation by single cell RNA-Seq.

FIG. 8A: iPAGE gene ontology/pathway analysis based on rank-ordering of

differentially expressed genes using +/-(l-p) where p is the p-value for differential expression between the U87- and MCFlOa-exclusive lanes given by the Wilcoxin rank- sum test. Values are positive for genes more highly expressed in U87 and negative for genes more highly expressed in MCFlOa. FIG. 8B: t-SNE clustering of 396 single cell profiles based on the differentially expressed genes color-coated by the lane-of-origin of each profile. Two clear spatial clusters form and each is predominantly associated with a specific cell type-exclusive lane. FIG. 8C shows the same t-SNE clustering shown in (FIG. 8B) but color-coated with a score indicating expression of the U87-specific genes vs. the MCFlOa-specific genes. The score is based on the relative rank ordering of U87- and MCFlOa-specific genes in each cell.

[0024] FIG. 9 provides for a schematic of a reversible PDMS-based device in one aspect of the disclosure.

[0025] FIG. 10 describes ingle Cell RNA-Seq of U87 and MCFlOa Cell Lines

-600 cells paired with beads in five-lane microfluidic device.

[0026] FIG. 11 shows the number of reads associated with each cell-identifying barcode showing a sharp change in slope at -350 barcodes, consistent with the -350 cell- bead pairs loaded in the microwell array. The remaining cell-identifying barcodes originate either from sequencing error or ambient, background mRNA hybridizing to beads in the device that were not associated with a cell. [0027] FIG. 12 is a histogram showing that the number of genes detected per individual cell ranges from 1,000 to 17,000 (-3,400 genes per cell on average).

[0028] FIG. 13A and FIG. 13B describes a representative example of the capabilities described herein for linking high content imaging to single cell sequencing. For example, introducing 4 different at 6 intensities generates 6 ⁴ or 1296 optical barcodes (FIG. 13A).

[0029] FIGS 14A and 14B provides for (FIG. 14A) a fluorescence image of beads with different fluorescent colors generated by combinatorial mixing of different fluorescent dyes at different concentrations. Each color combination and intensity represents a different sequence barcode for mRNA capture that is also attached to the bead. In this example the optical barcodes themselves are not sequences. FIG. 14B shows the intensity distribution for the beads in one of the three fluorescence channels demonstrating our ability to separate the optical barcodes based on fluorescence intensity (in this case with five different intensity levels).

DETAILED DESCRIPTION

[0030] In an aspect, the disclosure provides for a microwell array system, devices, and methods for pairing individual cells with mRNA capture beads. In another aspect, the disclosure provides for a microwell array system, device, and methods for pairing individual cells with mRNA capture beads that introduce, cell-identifying barcode sequences into cDNA generated after mRNA capture. In an aspect, the system is compatible with high-content cellular imaging and drug stimulation experiments.

[0031] The disclosure further provides for capture beads, for example a mRNA capture beads, or probes comprising one or more optical barcodes described herein. In an aspect, the capture beads, for example a mRNA capture beads, or probes are used to identify drug and/or to be used in drug simulation experiments.

[0032] In an aspect, the disclosure provides for devices, systems, and/or methods wherein cells are arranged in a grid and/or a chamber. In yet another aspect, the disclosure provides for devices, systems, and/or methods wherein cells are arranged in a grid and/or chamber as compared to devices, systems, and/or methods wherein cells are arranged in droplets or a configuration where cells are physically segregated from one another. In an aspect, a microfluidic device, system, or method described herein comprises a flow cell with an array of microwells embedded in either the top or bottom of the device. In another aspect, the device described herein is a solid state device that allows for single cell isolation, imaging, and/or uniform parallel introduction of reagents to a plurality of cells.

[0033] The disclosure provides for devices, systems, and/or methods wherein cells are arranged in a reversible chamber or microchamber wherein the chamber and/or microchamber can be open and/or closed. The disclosure also provides for devices, systems, and/or methods wherein cells are arranged in reversible chamber or

microchamber where the chamber and/or microchamber can be opened or closed as compared to devices, systems, and/or methods wherein cells are arranged in nonreversible droplets or a non-reversible configuration wherein cells are physically segregated from one another. In such a non-reversible configuration, the droplets and/ non-reversible configuration wherein cells are physically segregated cannot be readily toggled back and forth, that is, may not be opened or closed. As a result, in a nonreversible configuration, fluidics cannot be uniformly distributed to the cells and then uniformly removed and replaced by a second fluid.

[0034] The disclosure further provides for devices or system comprising

(a) one or more capture beads, for example mRNA capture beads;

(b) one or more cell-identifying barcode sequences, for example, optical barcode sequences;

(c) a plurality of chambers or microchambers including one or more mRNA

capture beads and/or one or more cell-identifying barcode sequences;

(d) wherein the plurality of chambers or microchambers including one or more mRNA capture beads and/or one or more cell-identifying barcode sequences are reversible and are capable of being opened or closed more than a single time.

[0035] In an aspect, the disclosure provides for devices or system comprising

(a) one or more mRNA capture beads;

(b) one or more cell-identifying barcode sequences; (c) a plurality of chambers or microchambers including one or more mRNA capture beads and/or one or more cell-identifying barcode sequences;

(d) wherein the plurality of chambers or microchambers including one or more mRNA capture beads and/or one or more cell-identifying barcode sequences are reversible and are capable of being opened or closed more than a single time; and

(e) wherein the device or system does not comprise a droplet microfluidics device or does not include droplet microfluidic technology, nonoliter droplet technology, and/or droplet sequence, "Drop-seq," "Drop-seq single cell analysis" technology, and/or technology wherein cells are captured in "droplets."

[0036] The disclosure further provides for probes comprising one or more mRNA capture beads and one or more cell-identifying barcode sequences associated or coupled with the one or more mRNA capture beads.

[0037] The disclosure further provides for compositions comprising one or more mRNA capture beads and one or more cell-identifying barcode sequences associated or coupled with the one or more mRNA capture beads and one or more cells.

[0038] The disclosure further provides for methods of drug discovery, drug profiling, and/or drug testing comprising

(a) combining one or more mRNA capture beads with one or more cell- identifying barcode sequences;

(b) adding one or more mRNA capture beads with one or more cell-identifying barcode sequences to a plurality of chambers or microchambers;

(c) wherein the plurality of chambers or microchambers including one or more mRNA capture beads and one or more cell-identifying barcode sequences are reversible; and

(d) adding one or more drugs to the reversible chambers or reversible

microchambers.

[0039] The disclosure further provides for a probe comprising

(a) one or more mRNA capture beads; (b) one or more cell-identifying barcode sequences, for example, optical barcode sequences;

(c) a plurality of chambers or microchambers including one or more mRNA

capture beads and/or one or more cell-identifying barcode sequences;

(d) wherein the plurality of chambers or microchambers including one or more mRNA capture beads and/or one or more cell-identifying barcode sequences are reversible and are capable of being opened or closed more than a single time

[0040] The disclosure further provides for methods for single cell RNA capture and sequencing, wherein said method comprises

(a) introducing at least one cell into a microwell, wherein the microwell is

attached to a first substrate that faces a second substrate and wherein oligo primers are attached to the surface of said second substrate;

(b) hybridizing mRNA molecules to the surface of said second substrate;

(c) adding at least one buffer to the microwell;

(d) contacting said first substrate with said second substrate, thereby creating a seal between said first substrate and said second substrate.

[0041] In an aspect, the device, system, and/or microarray is imaged using florescence.

[0042] In another aspect, the disclosure provides for a device for single cell RNA capture and sequencing, wherein said device comprises a plurality of microwells; wherein said plurality of microwells are attached to a first polysiloxane substrate; a second substrate comprising glass that faces said first substrate, wherein oligo primers are grafted onto to the glass surface of said second substrate.

[0043] In an aspect, the substrate or surface comprises a polysiloxane substrate.

In an aspect, the polysiloxane substrate comprises PDMS. In another aspect, the substrate comprises polymethylmethyacrylate (PMMA). In another aspect, the substrate comprises a thermoplastic for fabrication by, for example, hot embossing. In another aspect, the substrate comprises glass for fabrication by, for example laser machining. In another aspect, the substrate comprises silicon or silicon-on-glass or photoresist-on-silicon. [0044] In another aspect, a system or device described herein further comprises one or more drugs, buffers, active agents, and/or a plurality of cells.

[0045] In an aspect, the disclosure provides for devices, systems, methods, and/or probes capable of profiling as hundreds or thousands of cells from an organ or tumor. In an aspect, the cells analyzed are healthy human cells, abnormal cells, cancer cells, neural cells, immune cells, epithelial cells, mesenchymal cells, or stem cells. Cells could also originate from microorganisms including parasites, fungi, or bacteria. Other units containing nucleic acids such as viruses both in isolation or inside of an infected host cell could be analyzed.

[0046] In yet another aspect, different drugs or chemical constituents or compositions, for example, lysis buffer, may be introduced to a device or system described herein, for example, in a chamber or microchamber, at any time given the reversible nature of the chambers or microchambers. This provides for additional flexibility and advantages over other devices or systems, such as those employing, for example, droplet microfluidic technology, nanoliter droplet technology, and/or droplet sequence, "Drop-seq," "Drop-seq single cell analysis" technology, and/or technology where cells are captured in "droplets."

[0047] In an aspect, the device or system described herein, for example, the microfluidic device does not involve any on-chip valves and/or moving parts, which can result in a high feature density. In another aspect, the device or system described herein is not a droplet microfluidics device or does not include droplet microfluidic technology, nonoliter droplet technology, and/or droplet sequence, "Drop-seq," "Drop-seq single cell analysis" technology, and/or technology where cells are captured in "droplets." In an aspect, the device or system described herein is more compatible with cell culture and cell perturbation assays than a droplet microfluidic device because fluids can be readily exchanged in a uniform fashion and cells can communicate with each other via diffusible factors. Hence, cell viability will be higher and cells will exhibit normal physiology in comparison to cells that are sequestered in droplets. In another aspect, in the device or system described herein, cell are arranged in regular array and are in contact with a flat, optically transparent surface as opposed to a droplet microfluidic device in which cells are isolated inside of a spherical enclosure. Therefore, in yet another aspect, the device or system described herein is intrinsically more compatible with imaging, for example optical imaging or microscopy, than droplet devices.

[0048] In an aspect, devices, systems, and methods described herein have a higher throughput than devices, systems, or methods employing droplet microfluidic technology, nonoliter droplet technology, and/or droplet sequence, "Drop-seq," "Drop-seq single cell analysis" technology, and/or technology where cells are captured in "droplets." In yet another aspect, analysis of cells, probes, and beads using devices, systems, and methods described herein have a throughput that is 5, 10, 15, 20, 25, or 100 or more times less expensive than devices, systems, or methods employing droplet microfluidic technology, nonoliter droplet technology, and/or droplet sequence, "Drop-seq," "Drop-seq single cell analysis" technology, and/or technology were cells are captured in "droplets."

[0049] In another aspect, the device described herein is compatible with automation, for example, a computer-controlled system that is capable of facilitating automated introduction of any of various fluids to the device, temperature control, and reversible sealing of the microwells. Accordingly, in a preferred aspect the device described herein can capture mRNA from thousands of individual cells in parallel and produce pooled, single cell RNA-Seq libraries for ~$0.10/cell.

[0050] In an aspect, barcoded capture beads are introduced to the microwell array in either a random process or ordered process. In yet another aspect, barcoded capture beads are introduced into the microwell array in a random process with little or no control over which barcoded capture bead enter each microwell. Under such a scenario, in an aspect, imaging information acquired from the cells in the device is not directly associated with the single cell transcriptomes that ultimately are quantified by deep sequencing. In an aspect, this issue is addressed by introducing barcodes, for example, optical barcodes, to the mRNA capture beads that indicate the sequence barcode attached to each bead. In another aspect, the optical barcodes can be read out using the same or different fluorescence microscope used for live cell imaging on the device. In yet another aspect, the disclosure provides for a reagent that allows for the capture RNA from individual cells, allows for association of a unique sequence barcode with each single cell cDNA library, and/or reveals the identity of the attached sequence barcode by optical methods. [0051] In an aspect, the disclosure provides for an optical barcode and associated methods described herein._In an aspect, the optical barcode is a sequence. In another aspect, the optical barcode is not a sequence. In another aspect, optical barcodes described herein are not DNA sequences.

[0052] In an aspect, an optical barcode described herein comprises a combination of fluorescent dyes comprising at least one different color and optionally more than one different intensity that make the bead uniquely identifiable on an optical microscope. In another aspect, an optical barcode described herein comprises a combination of fluorescent dyes comprising two, three, four, five or more different colors and optionally more than one different intensity that make a respective bead uniquely identifiable on an optical microscope

[0053] The disclosure further provides for proteomic methods and systems utilizing a device, probe, or bead described herein. In an aspect, an oligonucleotide-based optical barcode is loaded onto beads that also harbor reactive groups for capturing protein (e.g. amine-, carboxyl-, or thiol-reactive groups). In yet another aspect, optical barcodes are loaded onto beads along with one or more antibodies capable of capturing specific proteins. Proteins can be quantified by fluorescence methods (e.g. fluorescently labeled antibodies) or mass spectrometry. In a further aspect, the disclosure provides for the use of DNA-labeled antibodies that are amplified, identified, and quantified by deep sequencing.

[0054] The devices, systems, and methods described herein provide for key distinctions and advantages over other devices, systems, and methods. For example, the microwell array device is constructed in such a way that it can be reversibly sealed during cell lysis and RNA capture. Significant loss of RNA occurs in our arrays when cells are lysed in unsealed or even imperfectly sealed arrays due to rapid diffusion of RNA molecules. In the context of the bead capture and RNA-Seq experiments, this could result not only in reduced RNA capture, but also significant cross-talk. Herein, advantage is taken of the physical properties of PDMS, namely its flexibility and hydrophobicity, for high-fidelity, reversible sealing which is difficult to achieve using the agarose hydrogel device reported previously. In addition, the devices, systems, and methods described herein demonstrate genome-wide single cell RNA-Seq. Additionally, the single cell capture and pooled library preparation scheme described herein costs $0.10-$0.20/cell even at a relatively modest scale of several hundred cells per run (see, for example, Table 7), compared to the <$l/cell estimated at the 10,000-cell scale for alternative approach, for example, one described in Fan HC, Fu GK, Fodor SPA: Combinatorial labeling of single cells for gene expression cytometry. Science 2015, 347:628-636, which is herein incorporated by reference in its entirety.

[0055] In an aspect, the disclosure provides for optical demultiplexing or optical barcoding of sequence-barcoded capture beads by, for example, attaching a unique combination of short oligonucleotides to each bead. In an aspect, while the sequence barcode associated with the RNA capture primer can be read out by DNA sequencing, the optical barcode can be read out on a standard fluorescence microscope. In another aspect, the oligonucleotide combination is unique to each sequence barcode and can be identified by hybridizing fluorescently labeled complementary oligonucleotides to beads. The presence or absence of a series of fluorescently labeled oligonucleotides on each bead after hybridization indicates the unique combination of short oligonucleotides attached to each bead and therefore the sequence barcode attached to each bead. Once the sequence barcode attached to each bead is identified, single cell transcriptome captured on each bead can be identified and read out using a sequencer with phenotypic information obtained from imaging the cells associated with each bead in each microwell of our device.

[0056] In an aspect, optical barcoding technology described herein is combined with applications beyond single cell RNA-Seq. In an aspect, an extension of the invention is bulk RNA-Seq or RNA-Seq of small numbers of cells. These can be seeded in the microwells or propagated from a single cell seeded in each microwell. The disclosure further provides for single cell DNA-Seq. The disclosure further provides for barcoded capture oligonucleotides that contain oligo(dT) for capturing mRNA from individual eukaryotic cells. Alternatively, the beads are functionalized with primers that are specific to targeted DNA loci or RNA transcripts. In yet another aspect, the beads are

functionalized with primers that have a random sequence and can therefore capture any DNA or RNA sequence from an individual cell. In an aspect, in each case, a

corresponding optical barcode is used to link imaging data acquired for a target cell or group of cells and sequencing data acquired from whatever nucleic acids are captured on the bead.

[0057] The disclosure further provides for proteomic methods and systems utilizing a device described herein. In an aspect, an oligonucleotide-based optical barcode is loaded onto beads that also harbor reactive groups for capturing protein (e.g. amine-, carboxyl-, or thiol-reactive groups). In yet another aspect, optical barcodes are loaded onto beads along with one or more antibodies capable of capturing specific proteins. Proteins can be quantified by fluorescence methods (e.g. fluorescently labeled antibodies) or mass spectrometry. In a further aspect, the disclosure provides for the use of DNA- labeled antibodies that are amplified, identified, and quantified by deep sequencing.

[0058] Single cell analysis is important for understanding how cells respond to drugs and other perturbations because phenotypic responses are inherently asynchronous and cells

PDMS Microwell Flow Cell for Single Cell Transcriptome Capture

[0059] In an aspect, the microfluidic platform described herein comprises a flow cell with an array of microwells embedded in either the top or bottom of the device. In an aspect, the device or system described herein comprises a high-throughput DNA sequencing and digital PCR device, as described in, for example, White AK,

Vanlnsberghe M, Petriv OI, Hamidi M, Sikorski D, Marra MA, Piret J, Aparicio S, Hansen CL: High- throughput microfluidic single-cell RT-qPCR. Proceedings of the National Academy of Sciences of the United States of America 2011, 108(34): 13999- 14004 or Men Y, Fu Y, Chen Z, Sims PA, Greenleaf WJ, Huang Y: Digital polymerase chain reaction in an array of femtoliter polydimethylsiloxane microreactors. Anal Chem 2012, 84(10):4262-4266, the contents of which are incorporated by reference herein in their entirety.

[0060] In an aspect, fluids can be driven through the flow cell manually by, for example, laminar flow using a syringe or pipette. Fluid exchange in the microwells occurs by diffusion, while cells and beads can be loaded by gravity. The microwell arrays may be fabricated from, a polysiloxane substrate, for example, polydimethylsiloxane (PDMS), a silicone rubber commonly used in soft lithography. PDMS allows inexpensive, rapid, and repeatable fabrication from molds produced on silicon in photoresist using standard photolithography. See, for example, McDonald JC, Whitesides GM:

Poly(dimethylsiloxane) as a material for fabricating microfluidic devices. Accounts of chemical research 2002, 35(7):491-499 and First Single Cell Expression Analysis with PDMS, Marcus et al., Analytical Chemistry, 2006. (Quake Lab), which are herein incoprtated by reference in their entirety. In addition, the material properties of PDMS, including its hydrophobicity and flexibility, facilitate reversible sealing of the microwells against a flat surface using mechanical deformation and negative pressure (see, for example, FIG. 2A) or introduction of oil by laminar flow (See, for example, FIG. 3A).

[0061] In an aspect, the device or system described herein is capable of solid- phase capture of RNA from individual cells via two modes of operation- RNA "printing" on glass and RNA capture on beads. For example, in an aspect, in RNA printing mode, individual cells are loaded in the microwells, which are fabricated in a PDMS slab that faces a glass coverslip. Oligo(dT) primers are covalently grafted to the glass surface so that mature mRNA molecules can be immobilized by hybridization of their poly(A) tails. Following the introduction of lysis buffer, the microwells can be sealed by mechanically placing them in conformal contact with the functionalized glass surface. Cell lysis releases mRNA into the solution confined by the microwells, resulting in hybridization to the oligo(dT)-coated glass coverslip. By placing the flow channel under negative pressure, the seal can be maintained in the absence of mechanical force, making the device transportable and readily accessible to an optical microscope. See, for example, White AK, Vanlnsberghe M, Petriv OI, Hamidi M, Sikorski D, Marra MA, Piret J, Aparicio S, Hansen CL: High- throughput microfluidic single-cell RT-qPCR. Proceedings of the National Academy of Sciences of the United States of America 2011,

108(34): 13999-14004, which is herein incorporated by reference in its entirety.

[0062] Because the mRNA is immobilized on a glass surface, enzymatic processing steps can take place on-chip, simply by sequential flow of reagents through the device. After incubating the trapped, single cell lysates with the glass capture surface, the seal may be released and the flow cell can be rinsed with a detergent-containing buffer followed by a reaction mixture containing DNase. Because the oligo(dT) primers are comprised of locked nucleic acid (LNA), they are resistant to nuclease digestion. See, for example, Koshkin AA, Nielsen P, Meldgaard M, Rajwanshi VK, Singh SK, Wengel J: LNA (locked nucleic acid): an RNA mimic forming exceedingly stable LNA: LNA duplexes. Journal of the American Chemical Society 1998, 120(50): 13252-13253, the content of which is incoroprated by reference in its entirety.

[0063] The immobilized single cell mRNA libraries can then be reverse transcribed in parallel, and the resulting mRNA/cDNA hybrids can be visualized by fluorescence microscopy after staining with a fluorogenic intercalator dye. For example, FIG. 4B illustrates a fluorescence image of single cell transcriptome "prints" arrayed on a glass coverslip as described above. Here, it is further confirmed by running a control experiment that the printed material originates from RNA. While some aggregates of genomic DNA were not fully digested (but are reduced in intensity by DNase treatment), the disclosure further confirms that the vast majority of material imaged in the circular prints originates from RNA. For example, the left-most panel of FIG. 4C shows a bright field image of a microwell array in which three microwells each contain an individual cell. The resulting RNA prints (middle panel) that can be visualized after reverse transcription are ablated by incubating the surface with RNaseH (right- most panel), which selectively digests RNA in RNA/DNA hybrids. Conversion of RNA/DNA hybrids to single-stranded cDNA precludes detection using the intercalator dye, and so removal of RNA from the prints eliminates the fluorescence signal almost completely. FIG. 4C also contains some small fluorescent objects associated with the interstitial walls of the microwell array or with microwells that did not contain a cell. These are substantially reduced in intensity by RNase treatment, confirming that they are, in fact, RNA that is spuriously captured or non-specifically absorbed. These objects could arise due to contamination from dead cells or other sources of freely floating RNA introduced with the cells prior to sealing. Nonetheless, the vast majority of the observed signal in FIG. 4C is associated with the circular mRNA prints that correlate perfectly with microwells that initially contained a cell.

[0064] FIG. 4A shows a second, very similar version of the device where the microwells are fabricated in PDMS on a glass slide, and the sealing is accomplished by laminar flow of oil. Using nearly the same procedures as described above for RNA printing mode, we use this version of the device to capture RNA on beads. After introducing cells, we can load beads into the microwells by gravity and achieve super- Poisson loading by using beads with a mean diameter greater than the radius of the microwells. Like the glass surface in FIG. 4A, we coat the beads in oligo(dT) to facilitate mRNA capture after cell lysis and sealing. FIG. 2B shows bright field and fluorescence images of a bead-containing microwell array loaded with individual cells following solid- phase mRNA capture and reverse transcription. The bead contained in a microwell that also contains a cell is substantially more fluorescent following reverse transcription than the other beads. While there is some fluorescence signal associated with beads that do not contain a cell, this is mainly due to non-specific staining of the high-density of single- stranded primers on the bead surface and non-specific staining of the bead itself, as shown in the third panel of FIG. 2B where we depict fluorescence images of beads in the absence of cells, cell lysate, or RNA as a negative control.

A Scalable Platform for Single Cell RNA-Seq

[0065] A scalable platform for single cell RNA-Seq based on the bead capture modality is described herein. The low reagent volumes required for microfluidic processing result in a significant cost reduction relative to some conventional methods, for example, as described in Wu AR, Neff NF, Kalisky T, Dalerba P, Treutlein B, Rothenberg ME, Mburu FM, Mantalas GL, Sim S, Clarke MF et ah Quantitative assessment of single-cell RNA- sequencing methods. Nat Methods 2014, l l(l):41-46, which is herein incorporated by reference in its entirety.

[0066] A further reduction in cost can be realized by, for example, using microfluidics in combination with schemes for cDNA barcoding. A representative example of this, such as the CEL-Seq strategy, can be found in Hashimshony T, Wagner F, Sher N, Yanai I: CEL-Seq: single-cell RNA-Seq by multiplexed linear amplification. Cell Rep 2012, 2(3):666-673, which is herein incorporated by reference in its entirety. By introducing a cell-specific barcode to the cDNA during reverse transcription, all subsequent sequencing library preparation steps can be accomplished on pooled cDNA from multiple cells, further reducing hands-on time and reagent consumption.

[0067] A pool of mRNA capture beads was generated in which each bead is attached to ~1 billion copies of a primer terminated on the 3 '-end with one of 960 possible barcode sequences followed by oligo(dT) using a combinatorial synthesis technique (FIGS. 3A and 3B). If, for example, 100 cells loaded into the micro wells of a device or system described herein receive a random barcoded bead from the pool, it is expected that the mRNA from -95 of them would be uniquely labeled based on the binomial distribution. A copy of the T7 promoter sequence (TPS) and part of an Illumina sequencing adapter (ISA) comprise the 5 '-end of the capture primer (Table 1) to allow linear pre- amplification by in vitro transcription (IVT) and library enrichment by PCR (FIG. 3B). To create this large pool of barcoded beads, 96 different barcode-containing oligonucleotides (Table 2) were copies onto a dual-biotinylated oligonucleotide containing TPS and ISA by primer extension with DNA polymerase in a 96- well plate. Each barcode is terminated with a universal, 6-base anchor sequence that becomes the 3'- end of the biotinylated oligonucleotide after the first round of primer extension (FIG. 3B). After this first reaction, each barcoded oligonucleotide was immobilized onto a set of streptavidin-coated Sepharose beads, quench the reaction, combine all of the barcoded beads in a pool, and remove original barcode-containing strand by denaturation. At this point, the pool of beads is split into ten new reactions and each containing one of ten unique second barcodes along with poly(dT) (Table 3) are added to the 3 '-end of the immobilized oligonucleotide by primer extension from the universal anchor sequence (FIG. 3B). After quenching this reaction, the beads can be pooled, removing the unbiotinylated strands, and washed. The resulting pool of beads contains 960 barcoded capture primers.

[0068] A PDMS microwell device containing five flow channel lanes for physical multiplexing of samples and > 10,000 microwells was constructed (FIG. 5A). The cylindrical microwells are about 50 μιη in diameter and height with a volume of <100 pL. In another aspect, the microwells are about 100 μιη to about 5 mm, about 5 μιη to about 200 μιη, about 10 μιη to about 100 μιη, about 25 μιη to about 75 μιη, about 30 μιη to about 600 μιη in diameter and height. In yet another aspect, the microwells comprise a volume of about 10 pL to about pL, about 10 pL to about 100 pL, about 25 pL to about 75 pL, about 25 pL to about 100 pL. In an aspect, cells are loaded in individual microwells randomly, according to Poisson statistics, such that the majority of cell-containing wells contain one cell. The concentration of the cellular suspension can be tuned in order to avoid overloading the microwell array. For example, if -100 cells are captured in every 1,000 microwells of a given array, then about <5% of microwells will contain more than one cell. Beads can then be loaded into the wells at a somewhat higher density because the mean diameter of the beads (-30 μιη) significantly reduces the probability double- loading (FIG. 5B and FIG. 5C). In an aspect, depending on the size of the microwell, it is rare to observe both beads and cells in an overloaded microwell. Given the pool of 960 cell-identifying barcodes and five lanes, the capacity of the system described herein for single cell RNA-Seq is -600 cells at a unique barcoding rate of >94%. In another aspect, the capacity of the system described herein for single cell RNA-Seq is -600 cells at a unique barcoding rate of about greater than 85%, greater than 90%, greater than 91%, greater than 92%, greater than 93%, greater than 94%, greater than 95% or more. In an aspect, the device or systems described herein may be scaled to increase capacity by synthesizing additional barcodes and/or adding microwells to the device or system.

[0069] After loading the cells and barcoded beads, procedures described above and herein can be employed to trap single cell lysates in sealed microwells, immobilizing captured mRNA on beads, and reverse transcribe (FIG. 5B). Following on-chip second- strand synthesis, the pool of single cell libraries can be simultaneously elute and pre- amplify overnight by IVT using T7 RNA polymerase (FIB. 5B). The resulting amplified RNA (aRNA) can be removed from each lane using a pipette, reverse transcribe the aRNA from each lane with primers containing lane-identifying barcodes, the cDNA libraries from all five lanes can be pooled, and enrichment of the sequencing library in a single PCR reaction can be undertaken. The primers used for aRNA reverse transcription contain 8-base unique molecular identifiers (UMIs) so that the vast majority of cDNA molecules are distinguishable. That way, genes can be quantified from sequencing data based on the number of UMIs associated with each gene rather than the number of reads, mitigating noise and bias that result from exponential amplification by PCR. See, for example, Shiroguchi K, Jia TZ, Sims PA, Xie XS: Digital RNA sequencing minimizes sequence-dependent bias and amplification noise with optimized single-molecule barcodes. Proceedings of the National Academy of Sciences 2012, 109(4): 1347-1352 and Shiroguchi K, Jia TZ, Sims PA, Xie XS: Digital RNA sequencing minimizes sequence- dependent bias and amplification noise with optimized single-molecule barcodes. Proceedings of the National Academy of Sciences 2012, 109(4): 1347-1352, which are incorpoaretd by reference in their entirety.

Demonstration and Analysis of Highly Multiplexed Single Cell RNA-Seq

[0070] In an aspect, the microfluidic device described herein is used to obtain

RNA-Seq profiles from -600 cells across five lanes from two commonly used human cancer cell lines (described herein as "Experiment 1" throughout the text). One of the five lanes contained U87 human glioma cells, one contained MCFlOa human breast cancer cells, and the other three contained a mixture of both cell lines. These two cell lines are highly mesenchymal, have been cultured for numerous passages, and have relatively similar expression profiles. Nonetheless, they are distinguishable by a few key genes and can be readily separated the data set. In addition, in an aspect, a slightly different protocol with less expensive reagents are used to obtain profiles of -500 cells across five lanes for a different cell pair (U87 cells and the diploid cell line WI-38, which has not undergone malignant transformation) in Experiment 2.

[0071] A factor with any pooled library scheme is cross-talk between cell- identifying barcodes. As described herein, this issue is addressed by quantitative analysis of Experiment 1. For example, the disclosure provides for methodology wherein crosstalk is quantified by using both sequencing and imaging data. In an aspect, because the device or system described herein is compatible with fluorescence microscopy, a fraction of the streptavidin molecules on each bead can be labelled with red-fluorescent

AlexaFluor 647 and pre-stained the cells with a blue-excitable live stain. This allows quantification of the number of cells paired with a barcoded capture bead and also allows for the estimation of the number of barcodes that are expected to be observed in the sequencing data for each lane. The sequencing data reveals that more barcodes are present in the library for each lane than expected based on the imaging data. Analysis reveals that the number of molecules associated with a given barcode placed the barcodes in two distinct populations. The size of the population of barcodes associated with a larger number of molecules was highly consistent with the imaging data (within -8%), which we take to demarcate our single cell RNA-Seq profiles. The second, larger population of barcodes with relatively few associated molecules likely results from multiple potential sources including sequencing error, actual cross-talk or spurious capture within our microfluidic device, and PCR jumping as observed in other implementations of multiplex single cell RNA-Seq. Across all five lanes the cell-identifying barcodes that were not associated with actual cells in the device had 200-300x fewer molecules per barcode than those associated with cells (based on the ratio of median unique molecules in the two populations).

[0072] As described in FIGS. 6A and 6B, the systems and devices described herein produce useful single cell RNA-Seq profiles. For example, in an aspect, the library preparation protocol described herein is based on CEL-Seq where, rather than sequencing the full gene body and normalizing by transcript length, the 3 '-end of transcripts are sequenced and counted. FIGS. 6A and 6B show the expected distribution of mapping positions for 3 '-end sequencing, with most reads mapping to the 3'-UTRs or coding sequences. Subsequent analysis to demonstrate cell type separation using our data set will rely on the 396 single cell profiles that we obtained with the highest coverage. Although 635 genes on average were detected across all cell profiled in Experiment 1, an average of 876 genes from the top 396 cells were detected (FIG. 6B). Hence, the 204 cells that we discard from subsequent analysis have an average of -170 genes detected per cell.

Similarly, for Experiment 2, an average of 1,030 genes from the top 247 single cell profiles were detected, but -530 genes on average across all cells.

[0073] To assess the similarity of our single cell expression profiles to conventional, population-level RNA-Seq, the Pearson correlation between bulk RNA-Seq and single-cell medians constructed from different numbers of individual cells after normalizing by the total number of molecules detected in each cell were calculated (FIG. 7A and FIG. 7B). An analysis on single cell profiles originating from the U87-exclusive and MCFlOa-exclusive lanes in Experiment 1 was conducted, randomly sampling the complete sets of profiles ten times without replacement for each point in the curves shown in FIG. 7A and FIG. 7B. This analysis shows that the single cell medians constructed from U87 single cell profiles correlate better with the bulk U87 RNA-Seq profile than with the bulk MCFlOa RNA-Seq profile (FIG. 7A), and vice- versa (FIG. 7B). It also shows that the single cell median correlations saturate around r=0.55-0.60 depending on the cell type. As a point of comparison, an analysis for CEL-Seq and DR- Seq gave population-level Pearson correlations of 0.71 and 0.69, respectively. See, for example, Dey SS, Kester L, Spanjaard B, Bienko M, van Oudenaarden A: Integrated genome and transcriptome sequencing of the same cell. Nat Biotechnol 2015, 33(3):285- 289, which is herein incorporated by reference in its entirety.

[0074] To further demonstrate the robustness of our data set, classifier for U87 and MCFlOa cells in Experiment 1 were built. Single cell profiles from lanes that contained either exclusively U87 cells of MCFlOa cells were used to identify 189 differentially expressed genes (p<0.05, Wilcoxin rank-sum test). FIG. 7C shows the log- ratio of the coefficients of variation (CVs) for each of these two genes sets between the mixed lane profiles and the profiles from the respective pure lanes. The log-ratio of CVs is greater than zero (CV ratio greater than one) for 92% of U87-specific genes and 85% of MCFlOa-specific genes.

[0075] FIG. 8A shows a pathway analysis of gene ontologies enriched across

> 11,600 genes that were both detected across our 396 single cell profiles and available in the iPAGE database and ranked based on differential expression in Experiment 1. See, for example, Goodarzi H, Elemento O, Tavazoie S: Revealing global regulatory perturbations across human cancers. Mol Cell 2009, 36(5):900-911, which is herein incorporated by reference in its entirety. A matrix of Spearman correlation coefficients across our 396 profiles based on rank-ordering the 189 differentially expressed genes in each cell was then generated. The data was clustered spatially using the t-stochastic neighborhood embedding (t-SNE) algorithm, described for example in Van der Maaten L, Hinton G: Visualizing data using t-SNE. Journal of Machine Learning Research 2008, 9(2579-2605):85, a clustering algorithm applied to high-dimensional single cell analysis data (FIG. 8B and FIG. 8C). This t-SNE result contains two closely associated clusters of individual cells. To understand the origin of these two clusters, the t-SNE clustering with two different color-schemes are displayed. In FIG. 8B, it is shown how single cell profiles from the various lanes of a device described herein are distributed. One of the two clusters contains all of the cells from the MCFlOa-exclusive lane, while the other contains nearly all of the cells from the U87 lane with a few exceptions. Single cell profiles from the mixed lanes are distributed throughout the two clusters. While the single cell profiles from mixed lanes are distributed uniformly throughout the MFC 10a cluster, there is some separation between a subset of mixed lane profiles and the U87- exclusive lane profiles in the U87 cluster. In FIG. 8C, the same clustering result with a different color scheme that indicates the relative rank ordering of the U87 vs. MCFlOa gene sets in each profile is described. This metric associates the cells in each of the two clusters with the expected cell type-specific expression pattern.

[0076] In an aspect, the microfluidic platforms described throughout are compatible with transcriptome-wide analysis of individual cells by RNA-Seq. In an another aspect, either could be combined with a sequence-based barcoding scheme to generate a pooled cDNA library from hundreds or thousands of individual cells. For example, in the bead capture device, barcoding is not strictly necessary because physical means could be used to extract the beads from the microwells for downstream processing with conventional labware. Alternatively, fluorescently labeled oligonucleotide probes could be used to image captured RNA molecules similar to RNA-FISH. Probes could be introduced sequentially, imaged, and removed in cycles or combined with previously reported multiplexing schemes. Similarly, sequential rounds of qRT-PCR in sealed microwells could allow targeted detection of specific genes or mutations in captured RNA.

Kits

[0077] In another aspect, the disclosure provides for a kit comprising, consisting essentially of, or consisting of any of probes, beads, and regents disclosed herein. In an aspect, the kit includes any of the combination of probes, beads, and regents disclosed herein in Figures 1 - 13. In another aspect, the kit provides for probes, beads, and regents applied in a manner that is consistent with the methodology of the examples and figures. In another aspect, the kit provides instructions or guidance regarding the use of the compositions, probes, devices, systems or methods described herein.

[0078] In an aspect, the kit includes instructions describing the methodology described herein.

[0079] The following examples serve to illustrate certain aspects of the disclosure and are not intended to limit the disclosure. EXAMPLES

[0080] The invention is illustrated herein by the experiments described by the following examples, which should not be construed as limiting. The contents of all references, pending patent applications and published patents, cited throughout this application are hereby expressly incorporated by reference. Those skilled in the art will understand that this invention may be embodied in many different forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will fully convey the invention to those skilled in the art. Many modifications and other embodiments of the invention will come to mind in one skilled in the art to which this invention pertains having the benefit of the teachings presented in the foregoing description. Although specific terms are employed, they are used as in the art unless otherwise indicated.

Example 1 :

Generation of Optically Barcoded mRNA Capture Beads for Optical Demultiplexing by Multi-Color Fluorescence Imaging for Single Cell RNA-Seq

N-succinimide-coated Sepharose beads with a mean diameter of -30 μιη were obtained from GE Healthcare in isopropanol. The beads were washed three times with water by centrifugation and split into 125 different tubes. The beads were then re-suspended in a reaction mixture with a final concentration of 100 mM sodium borate (pH 8.5) along with three differently-labeled streptavidin proteins (streptavidin-AlexaFluor 488, streptavidin- AlexaFluor 546, and Streptavidin-AlexaFluor 647 from Life Technologies). For each of the 125 reaction mixtures, the three proteins were added at one of 5 concentrations such that each of the 125 reaction mixtures had one of 5 ^A 3 = 125 unique combinations of labels and label intensities. The reaction was incubated at room temperature for one hour on a rotisserie to allow the labeled streptavidins to covalently react with the beads, form amide bonds. The beads were then washed five times in 50 mM Tris-HCl pH 8, 50 mM NaCl, and 0.1% Tween-20 before using to completely quench any remaining reactive groups on the beads. Each of the 125 uniquely labeled sets of beads were then incubated with a uniquely barcoded, 5'-biotinylated oligonucleotide. The oligonucleotides had a universal adapter sequence on the 5 '-end to facilitate amplification, a barcode sequence, and oligo(dT) on the 3 '-end to facilitate mRNA capture. The oligonucleotides became bound to the beads through the biotin-streptavidin interaction, resulting in beads that could be identified by fluorescence microscopy based on their unique configuration of fluorescent labels to harbor a specific nucleic acid barcode sequence.

Generation of Optically Barcoded mRNA Capture Beads for Optical Demultiplexing by Sequential Fluorescence Hybridization for Single Cell RNA-Seq

[0081] 2.4 million beads aminated polystyrene beads (Custom Primer Support 200

Amino, GE) were reacted with an 8% solution of glutaraldehyde in Cyanoborohydride Coupling Buffer (Sigma) for six hours a room temperature. After washing the beads in phosphate buffered saline (PBS), the beads were divided evenly into each of 32 tubes. To each tube, 62.5 pmol of a unique 5 '-aminated oligonucleotide were added containing an adapter sequence, a unique capture barcode sequence, and a linker sequence. In addition, zero or more 5 '-aminated oligonucleotides were added from a set of five, each of which contains a unique optical barcode sequence and a linker sequence. The constellation of oligonucleotides on the bead surface forms a binary code. For example, if there are five oligonucleotides in a set, and the first, third, and fourth oligonucleotides are present on the bead, then the code could be in an aspect, 10110. However, there will also be a bead in the set with the code 00000, where none of the oligonucleotides are present. This also provides a signal, a bead that does not fluoresce when exposed to any of the probes

[0082] .2.5 pmol of each optical barcode oligonucleotide was further added. Each capture barcode sequence is associated with a unique set of optical barcode sequences. In this embodiment, there are 32 capture barcode sequences and 2 ⁵ =32 possible

combinations of optical barcode configurations added in the first round of synthesis. The 5 '-aminated oligonucleotides were reacted with the aldehyde-conjugated beads in Cyanoborohydride Coupling buffer for 12 hours at room temperature. The reaction was then quenched by the reaction by adding 1 M Tris-HCl to each well. The beads were then washed in PBS.

[0083] In a second barcode synthesis, the 32 reactions from the first round were pooled and then re-distributed into each of 128 tubes. To each tube, 150 pmol of a unique oligonucleotide was added containing a linker sequence complementary to that in the capture barcode oligonucleotides already attached to the beads, a second capture barcode sequence, a random sequence (for unique molecular identification), and a poly(A) sequence. In addition, zero or more oligonucleotides from a set of seven were added, each of which contains a unique optical barcode sequence and a linker sequence complementary to that in the optical barcode oligonucleotides already attached to the beads. 20 pmol of each optical barcode oligonucleotide. Each capture barcode sequence is associated with a unique set of optical barcode sequences. There are 128 capture barcode sequences and 2 ⁷ =128 possible combinations of optical barcode configurations added in the second round of synthesis. In combination with the first round of synthesis, the second round of synthesis results in a set of 32x128 = 4,096 uniquely barcoded beads with both unique capture and optical barcodes attached. The second round

oligonucleotides was hybridized to the oligonucleotides attached in the first round in Buffer 2 (New England BioLabs) with 1% tween-20 (Sigma) at 50C for 20 minutes and at room temperature for 12 hours. The beads were then washed and then copied the second round oligonucleotides onto the 3 '-end of the first round oligonucleotides by primer extension with Klenow Large Fragment DNA Polymerase (New England BioLabs) at room temperature for 2 hours. The primer extension reaction was quenched with EDTA followed by sodium hydroxide. The beads from all 128 reactions were then pooled and washed them extensively in sodium hydroxide and Wash Buffer (10 mM Tris pH 8, 1 mM EDTA, 0.01% tween-20).

Fabrication of a PDMS Microwell Array Device

[0084] Microfabricated arrays of cylindrical pillars in photoresist on a silicon wafer were obtained from FlowJEM. The pillars were 50 microns in diameter and 50 microns tall. A second wafer contained a relief pattern of a flow channel. A 1: 10 mixture of liquid polydimethylsiloxane (PDMS) base and curing agent (Sylgard 184, Dow- Corning) was poured on each wafer. A glass microscope slide was pressed onto the uncured liquid PDMS that had been poured on the wafer containing cylindrical pillars, forming a thin layer of PDMS between the wafer and the glass slide. Both wafers with liquid PDMS were then placed in an oven to cure at 90°C overnight. The cured, solidified PDMS were then peeled from both wafers resulting in a thin microwell array on the glass slide and a PDMS flow channel. After using a punch to insert inlet and outlet holes on either end of the flow channel, air plasma was used to bond the flow channel over the microwell array. The resulting microwell array flow cell device was then used for single cell RNA-Seq and optical demultiplexing.

Single Cell RNA-Seq Library Generation in a Microwell Array Device using In Vitro Transcription

[0085] Prior to the experiment, the device was flushed with 0.1% Tween-20 and incubated for several hours to hydrate the microwells, which were subsequently washed with PBS. Cell suspensions were counted using the Countess automated cell counter (Life Technologies). A suspension of cells in PBS mixed with Calcein AM (live stain) was flowed into the device and incubated for five minutes. Cells were deposited in the microwells by gravity. After thoroughly removing any excess cells with PBS, a suspension of barcoded capture beads that had been pre-counted by microscopy was introduced in PBS and allowed to load under gravity for five minutes. Excess beads were thoroughly removed with PBS and the flow cell was incubated on ice. A lysis buffer containing 0.08% Triton X-100 supplemented with SUPERaselN (Life Technologies) was flowed under ice-cold conditions immediately followed by fluorinert oil (Sigma) to seal the device. After two cycles of freeze-thaw at -80C to enhance cell lysis, the device was incubated at room temperature for 60 minutes for mRNA capture. After one hour of incubation for mRNA capture, the device was unsealed by rapid removal of the fluorinert oil with 20 mM Tris-HCl pH 8 containing 1% Triton X-100 and SUPERaselN followed by thorough washing of the device with 20 mM Tris-HCl pH 8, 50 mM NaCl, and 0.1% Tween-20.

[0086] The mRNA captured on the beads was reverse transcribed using

ProtoScript II Reverse Transcriptase (New England BioLabs) for two hours at 42C in lx ProtoScript Reverse Transcriptase Buffer supplemented with 10 mM DTT, 0.5 mM dNTPs, 0.1% Tween-20, and SUPERaselN. The reaction mixture was then removed with 20 mM Tris-HCl pH 8, 50 mM NaCl, and 0.1% Tween-20. Second strand synthesis was carried out using the MessageAmp II aRNA amplification kit (Ambion) according to the manufacturer's instructions. This involves a reaction mixture with DNA polymerase and RNaseH that is incubated for two hours at 16C. After removing the second strand reaction mixture with 20 mM Tris-HCl pH 8, 50 mM NaCl, and 0.1% Tween-20, in vitro transcription (IVT) was carried out using reagents in the MessageAmp II aRNA amplification kit (Ambion) including T7 RNA polymerase according to the

manufacturer's instructions for 13 hours at 37C. The reaction linearly amplified the double-stranded cDNA libraries on the beads, eluting pools of barcoded aRNA into the flow channel which was subsequently removed using a pipette and purified using an RNA Clean & Concentrator Column (Zymo). aRNA was then reverse transcribed using uniquely barcoded random hexamer primers using PrimeScript Reverse Transcriptase (Clontech) supplemented with 0.5 mM dNTPs, 10 mM DTT, lx PrimeScript Buffer, and 0.1 U/uL SUPERaselN. The reaction proceeded for 10 minutes at 25C and two hours at 42C. The RNA-cDNA hybrid product was purified twice using 0.65x Ampure XP beads (Beckman). The resulting cDNA libraries were then enriched by PCR using Phusion High Fidelity DNA polymerase (New England BioLabs) using the Illumina RP1 and RPI primers. The resulting PCR product was then purified on a 1.5% agarose gel, extracted using a Gel Extraction Kit (Qiagen), purified using 0.65x Ampure XP beads (Beckman), and sequenced on a NextSeq 500 sequencer (Illumina).

Single Cell RNA-Seq Library Generation in a Microwell Array Device using Template- Switching

[0087] Cells and mRNA capture beads were loaded into the microwell array device as described above. The device was then placed on a computerized temperature and fluidic control system for automated single cell library production. The system consists of a manually controlled pressurization system, software, an electronic six-port rotary selector valve (Rheodyne), and a thermoelectric heater-cooler module (TE

Technology). After cooling the system to 4C, a lysis solution comprised of TCL Buffer (Qiagen) and 1 % 2-mercaptoethanol was introduced to the device, which was rapidly and automatically sealed with fluorinert (Sigma). After removing the oil and unsealing the microwells, a reverse transcription reaction mixture comprised of lx Maxima RT Buffer, 1 mM dNTPs, 1 U/uL SUPERaselN, 2.5 uM template- switching oligonucleotide, 0.1% Tween-20, and 10 U/uL Maxima H- Reverse Transcriptase (ThermoFisher) was introduced and incubated for 30 minutes at room temperature followed by 90 minutes at 42C.

[0088] After removing the reverse transcription reaction mixture, the beads were removed from the microwells by sonication, concentrated by centrifugation and incubated for 30 minutes at 37C with Exonuclease I in lx Exonuclease I Buffer (New England BioLabs). The beads were then washed in 10 mM Tris-HCl pH 8, 1 mM EDTA, and 0.5% sodium dodecyl sulfate (SDS) followed by 10 mM Tris-HCl pH 8, 1 mM EDTA, and 0.01% Tween-20. The beads were then re-suspended in a PCR reaction mixture containing lx HiFi Hot Start Ready Mix (Kappa) and 1 uM PCR primer and

thermocycled to amplify a full-length, double- stranded, pooled cDNA library. The PCR product was purified with 0.6x Ampure XP beads (Beckman). The library was then fragmented and a sequencing adapter introduced using the Nextera XT library preparation kit (Illumina) according to the manufacturer's instructions. Finally, the sequencing library was enriched by PCR using a reverse PCR primer from the Nextera XT kit and a custom forward PCR primer along with amplification reagents from the Nextera XT kit used according to the manufacturer's instructions. The resulting PCR product was purified with 0.6x Ampure XP beads (Beckman) and sequenced on a NextSeq 500 sequencer (Illumina).

Optical Demultiplexing of Barcoded mRNA Capture Beads

[0089] Following cDNA synthesis on the surface of optically barcoded capture beads, the microwell array device was transferred to an epifluorescence microscope system. The microscope system contains an inverted microscope frame (Ti-U, Nikon), an electron-multiplied charge coupled device (EM-CCD) camera (iXon, Andor), a 20x 0.75 NA air objective lens (Nikon), a three-axis motorized stage (ASI), a 140 mW 637 nm optically pumped semiconductor laser (OBIS, Coherent), and custom optics for coupling the laser beam into the microscope. The custom optical path includes an engineered diffusor (Thorlabs) and electronic phase scrambling device for flattening the illumination field. An automated, computer-controlled fluidics system is attached the microscope system and contains two, ten-channel rotary selector valves (Rheodyne). A single computer program controls the fluidics system, laser shutter, camera, and microscope stage. Each of twelve, Cy5-labeled optical barcode probes was loaded into the channels of the fluidics system. The optical barcode probes were each complementary to one of the optical barcode oligonucleotides attached the beads as described above. In addition, a wash buffer (20 mM Tris-HCl pH 8, 50 mM NaCl, and 0.1% Tween-20) and a denaturing solution (100 mM NaOH) were loaded into independent channels of the fluidics system. The first of twelve optical barcode probes was flowed into the flow cell, incubated at room temperature for ten minutes, and washed out with wash buffer. After washing, the microscope automatically scanned and imaged the entire microwell array and identifies the beads in each microwell that are fluorescently labeled with the first optical barcode probe by hybridization, indicating the presence of the first optical barcode sequence. The denaturing solution was then introduced to the flow cell which causes the first optical barcode probe to dissociate from the beads and exit the flow cell. After washing with wash buffer, the process was repeated for all of the remaining optical barcode probes. At conclusion, the configuration of optical barcode sequences attached to each bead can be inferred from the series of twelve fluorescence images. Because each unique

configuration of optical barcode sequences corresponds to a specific capture barcode sequence associated with the cDNA that is also attached to each bead, the capture barcode sequence for each individual cell can be identified (see, for example, FIG. 1A and FIG. IB). This allows a direct association between imaging information acquired from the cells, the capture barcode associated with the cDNA of the at cell that is read out on the sequencer, and therefore the transcriptome of that cell all through identification of the corresponding optical barcode.

Example 2:

Fabrication PDMS Microwell Arrays for Single Cell RNA Printing

[0090] Silicon wafer masters (~4 in) with cylindrical pillars (diameter 50 micron; height 30 micron) for photolithography were obtained from Stanford Microfluidics Foundry and were subsequently exposed to lH,lH,2H,2H-perfluorooctyltrichlorosilane (Alfa Aesar) vapor under vacuum for -30 minutes to avoid curing of the PDMS on the wafer. PDMS (Sylgard 184, Dow Corning) was thoroughly mixed 10:1 (base:curing agent) and degassed under house vacuum for 2 hours. -15 g of degassed PDMS was poured onto the 4 in silicon wafer master and allowed to cure overnight at ~90°C. This slab with microwells was then gently peeled off from the master and used to construct PDMS microreactor flow cells.

Surface Chemistry on Glass Coverslip

[0091] Vista Vision Microscope cover glass (22x50x0.16 mm) was plasma sterilized (Harrick Plasma) for ~5 mins, and immediately immersed in 10% acetic acid (pH 3.5) ethanol solution containing 0.5% trimethoxysilanealdehyde (United Chemical Technologies), and incubated for 15 mins. The cover glass was then washed with ethanol, air-dried and heat cured at 90°C for 10 mins. A 2.5μΜ solution of 5 '-animated- LNA- oligo(dT) (Exiqon) in cyanoborohyride coupling buffer (Sigma) supplemented with 1M NaCl was added on the aldehyde surface of the cover glass. The cover glass was incubated for 3 hours at room temperature inside a humid chamber, and then washed with DI water. The aldehyde surface was then incubated in 10% ethanolamine in

cyanoborohydride coupling buffer for 30 mins to quench the unreacted aldehydes.

Construction of the Flowcell

[0092] A rectangular slab (3.5x1.5x0.1 cm) of PDMS containing the microwell array in the center was cut and a double-sided adhesive tape (-120 micron thickness, Grace BioLabs) was adhered to the flat side of the PDMS slab that contained the microwells. The tape was cut in an elongated hexagonal shape, which formed the microchannel in the flowcell. The other side of the tape was pasted on the LNA coated cover glass to build the microfluidic device. Two holes were punched at the two end of the microchannel with a biopsy punch, which acted as the inlet and outlet of the device and tubing were attached to allow liquid flow. The periphery of the PDMS slab was sealed on to the cover glass using epoxy glue.

Experimental Procedure for Single Cell mRNA printing on glass

[0093] A suspension of U87 cells in PBS was flowed in to the device and loaded into the microwells by gravity (kept upside down) for 5 mins at room temperature. After washing with PBS buffer supplemented with SUPERaseIN (Ambion), the microwells were sealed using an automated mechanical device by placing the flow cell upside down on a screw mounted on a motorized z-stage (ASI) so that the top PDMS slab containing the microwells was pressed against the glass bottom. After sealing the wells

mechanically, the seal was retained by hermetic sealing to trap the single cell lysate within a single microwell. The cells were lysed by freeze-thaw. Once the cells lysed, the mRNAs were captured on the LNA surface by hybridization of the 3'-polyA tail of the mRNA to the LNA-oligo(dT) during a 60 minute incubation. The microwells were then unsealed and the flow cell was immediately and vigorously washed with the Wash Buffer (20mM Tris pH 8.0, 50mM NaCl, 0.1% Tween-20), supplemented with SUPERaseIN (FIG. 2A). The flowcell was then incubated with TURBO DNase (Ambion) in TURBO DNase buffer, supplemented with 0.1% Tween-20 and SUPERaseIN for 30 mins at 37°C to digest any residual genomic DNA. The mRNA captured on the LNA surface was reverse transcribed using M-MuLV Reverse Transcriptase (New England Biolabs) for 2 hours at 42°C in lx M-MuLV Reverse Transcriptase buffer, supplemented with lOmM DTT, 5mM dNTPs, 0.1% Tween-20 and SUPERaseIN. After reverse transcription the double stranded RNA-cDNA hybrids were stained with ΙΟηΜ SYTOX Orange dye (Invitrogen), an intercalator that is selective for double- stranded DNA, and incubated for 5 mins prior to imaging.

[0094] The epifluorescence imaging system was constructed on an inverted Nikon

Eclipse Ti-U microscope with 20x, 0.75 NA air objective (Plan Apo λ, Nikon). SYTOX Orange was excited using a 532 nm diode -pumped solid state laser (Dragon Lasers), and the fluorescence was collected and imaged onto an electron multiplying charge coupled device (EMCCD) camera (iXON3, Andor Technologies). The images were acquired with 0.5 s exposure time (controlled by external shutter) at lMHz digitization (with no EM gain). Automated scanning of the surface (motorized X-Y stage, ASI), image acquisition, and illumination were controlled with custom software written in C/C++. The images were analyzed using ImageJ software.

Microfluidic Device for Single Cell RNA-Seq [0095] For the single cell RNA-Seq experiment a monolithic PDMS was designated based multi-channel device, by fabricating each channel with a microwell array. Two key soft lithography techniques were used to fabricate this device. First, instead of using silicon wafer master directly for fabricating the microwell array as done in the case of RNA printing device, we generated a secondary master made out of PDMS. This was done because the aspect ratio of the micropillars results in a relatively fragile silicon master. It was found that the PDMS master to be more durable. Second, instead of using a double-sided adhesive tape for the device assembly, the bottom and the top of the device were bonded together by partial curing. This provided us with more durable and reliable partitions between the individual channels of our device than could be generated using tape. For the multi-lane microfluidic device, two different silicon wafer masters were fabricated, one for the top and other for the bottom containing the array of microwells. Masters for soft lithography were generated from 4-inch silicon test wafers (University Wafer) coated with SU-8 2005 (MicroChem) photoresist. The wafer master for the bottom of the device contained five arrays of cylindrical pillars (diameter 50 micron; height 50 micron). The wafer was then fluorosilanized as described above. To avoid repeated use of the silicon wafer, we fabricated secondary masters in PDMS as follows. 40 g of degassed PDMS 10: 1 (base:curing agent) was poured and cured on the wafer, and then peeled off and cut into a rectangular slab. The surface containing an array of microwells was oxidized in plasma chamber (Harrick Plasma) for ~2 mins and immediately fluorosilanized. Using this microwell-containing slab as a master, -10 g of degassed PDMS was cured on it and peeled off. This new PDMS slab containing array of pillars is an exact replica of the silicon wafer, and is fluorosilanized and served as a secondary master for soft lithography for microfabrication of the bottom part of the microfluidic flowcell device. ~2 g of degassed PDMS 10: 1 (base : curing agent) was poured on a plasma cleaned glass slide and the secondary master with pillar array was placed gently with pillars immersed into the liquid PDMS. The slide, PDMS and master (on top) was degassed for ~5 mins and then cured hard at 90°C for 2 hours. After curing, the master is peeled off and a thin layer of PDMS is bonded to the glass slide with 5 lane arrays of microwells. [0096] A second silicon wafer master was constructed containing five longitudinal ridges (with a height of 100 microns) with rounded ends on which -30 g of degassed PDMS 15:1 (base : curing agent) was poured and allowed to cure partially at 60°C for 90 mins. The partially cured PDMS was cut into a slab, holes were punched at either end of each channel, and the slab was placed gently on the top of the glass slide containing the microwell array in such a way that the longitudinal grooves were aligned over each of the five microwell arrays. The slide assembly was then incubated at 90°C overnight to form a single monolithic PDMS structure as shown in FIG. 5A.

Synthesis of Uniquely Barcoded Beads for mRNA capture

[0097] N-succinimide-coated Sepharose beads with a mean diameter of -30 μιη were obtained from GE Healthcare in isopropanol. The beads were washed three times with water by centrifugation and re-suspended in a reaction mixture with a final concentration of 100 mM sodium borate (pH 8.5) and -0.8 mg/mL streptavidin

(streptavidin from New England BioLabs was spiked with -2% AlexaFluor 647-labeled streptavidin from Life Technologies). The reaction was incubated at room temperature for one hour on a rotisserie to allow the streptavidin to covalently attach to the beads. The beads were then washed five times in Wash Buffer and incubated in 1 Wash Buffer for 30 minutes before using to completely quench any remaining reactive groups on the beads.

[0098] A dual-biotinylated oligonucleotide containing both the T7 promoter sequence and a partial Illumina adapter sequence (Table 1) were annealed to each of 96 oligonucleotides (Table 2) that are complementary to the partial Illumina adapter sequence on the 3 '-end and contain a unique barcode and universal anchor sequence on the 5 '-end (FIGS. 3A and 3B). Table 1 describes a list of oligonucleotides used for barcoding and library preparation in additional to list of cell-identifying barcodes in Table 2 for Experiment 1.

Table 1 RT1 (Reverse Transcription GCCTTGGCACCCGAGAATTCCANNNNNNNNCGTGATNNNNNN Primer for Lane 1)

RT2 (Reverse Transcription GCCTTGGCACCCGAGAATTCCANNNNNNNNACATCGNNNNNN Primer for Lane 2)

RT3 (Reverse Transcription GCCTTGGCACCCGAGAATTCCANNNNNNNNGCCTAANNNNNN Primer for Lane 3)

RT4 (Reverse Transcription GCCTTGGCACCCGAGAATTCCANNNNNNNNTGGTCANNNNNN Primer for Lane 4)

RT5 (Reverse Transcription GCCTTGGCACCCGAGAATTCCANNNNNNNNCACTGTNNNNNN Primer for Lane 5)

RP1 (PCR Primer 1) AATGATACGGCGACCACCGAGATCTACACGTTCAGAGTTCTACAGTCC

GA

RPI1 (PCR Primer 2) CAAGCAGAAGACGGCATACGAGATCGTGATGTGACTGGAGTTCCTTGG

CACCCGAGAATTCCA

Table 2

FBC_01igo28 CAGGTCGGCTTATGAGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo29 CAGGTCCTAGTCCTCGATCGTCGGACTGTAGAACTCTGAAC

FBC_Oligo30 CAGGTCCTAGAGGAGGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo31 CAGGTCAGCTTTACCGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo32 CAGGTCGTCCATGAAGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo33 CAGGTCCTCGAACCTGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo34 CAGGTCCATTGTACGGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo35 CAGGTCTTGAACGCTGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo36 CAGGTCTACGTCATGGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo37 CAGGTCAAGCCGTTAGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo38 CAGGTCCGGACGTATGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo39 CAGGTCTCGTTACCGGATCGTCGGACTGTAGAACTCTGAAC

FBC_Oligo40 CAGGTCATCCCCCATGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo41 CAGGTCCAGACGATTGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo42 CAGGTCATCGATCCCGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo43 CAGGTCCCTGAGGATGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo44 CAGGTCAGCTCTTTGGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo45 CAGGTCGGAATACGGGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo46 CAGGTCCTATCCTGGGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo47 CAGGTCGGTTGTAGTGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo48 CAGGTCGAACGTAGCGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo49 CAGGTCGTCTATCGGATCGTCGGACTGTAGAACTCTGAAC

FBC_Oligo50 CAGGTCTACGAGTGGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo51 CAGGTCTCATGTCGGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo52 CAGGTCAAACACCCGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo53 CAGGTCACTAGTCCGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo54 CAGGTCCGAGGAATGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo55 CAGGTCACAATGGCGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo56 CAGGTCTAGGTCTCGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo57 CAGGTCTCTGTGAGGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo58 CAGGTCGGGATTGAGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo59 CAGGTCAACTCTGGGATCGTCGGACTGTAGAACTCTGAAC

FBC_Oligo60 CAGGTCAAACGCGTGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo61 CAGGTCTCCTACGAGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo62 CAGGTCTAGCAGGTGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo63 CAGGTCCCTGCATTGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo64 CAGGTCGTGATGCAGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo65 CAGGTCCGATTCAGGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo66 CAGGTCAGGATGACGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo67 CAGGTCAGGCCATAGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo68 CAGGTCGCTTGCTTGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo69 CAGGTCTCCCAAGTGATCGTCGGACTGTAGAACTCTGAAC

FBC_Oligo70 CAGGTCTCAAGGCAGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo71 CAGGTCACGAGGTAGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo72 CAGGTCGGAACGAAGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo73 CAGGTCAATCCCAGGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo74 CAGGTCCGATAAGGGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo75 CAGGTCTATCGCGAGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo76 CAGGTCCGCATAACGATCGTCGGACTGTAGAACTCTGAAC FBC_01igo77 CAGGTCGTGCAGTTGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo78 CAGGTCAGAACGCTGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo79 CAGGTCTAGAGGTCGATCGTCGGACTGTAGAACTCTGAAC

FBC_Oligo80 CAGGTCCTGTGATGGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo81 CAGGTCTAGAGCCAGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo82 CAGGTCCTTGATGCGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo83 CAGGTCTTCGTGTCGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo84 CAGGTCTATCTGCGGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo85 CAGGTCTGGTAGGAGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo86 CAGGTCCCTAGACAGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo87 CAGGTCAGTCAACGGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo88 CAGGTCAAGGGTGAGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo89 CAGGTCCTTCACACGATCGTCGGACTGTAGAACTCTGAAC

FBC_Oligo90 CAGGTCAGGTTGCTGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo91 CAGGTCACCCGAAAGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo92 CAGGTCGAAAAGGGGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo93 CAGGTCACTTCCCAGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo94 CAGGTCTGCTGCATGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo95 CAGGTCATTCCTGGGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo96 CAGGTCCAGAACTCGATCGTCGGACTGTAGAACTCTGAAC

[0099] Table 2 describes a list of oligonucleotide sequences used to generate the first set of barcoded beads (FBC) for combinatorial synthesis in Experiment 1.

[00100] The dual-biotinylated oligonucleotide was annealed at a final

concentration of 2 μΜ in the presence of a four- fold molar excess of the barcoded oligonucleotide in a 96-well plate by stepwise cooling from 85C to 30C over 30 minutes. A DNA polymerase master mix was then added to each well such that the final concentration of the reaction components was lx NEB Buffer 2 (New England BioLabs), 0.25 υ/μί Klenow Fragment (exo-) (New England BioLabs), and 0.5 mM dNTPs. The reaction was incubated in each well at 37C for 30 minutes before heat inactivating the polymerase at 75°C for 20 minutes.

[00101] An equal volume of beads was then added to each reaction mixture so that the extended, dual-biotinylated oligonucleotide could conjugate to the streptavidin coated beads at a final density of ~1 billion oligonucleotide primers per bead. The conjugation reaction was incubated at room temperature overnight on a rotisserie and quenched with biotin at a final concentration of 2 mM and sodium hydroxide at a final concentration of 125 mM to melt the template strand off of the beads. The beads were then pooled and washed five times in 125 mM sodium hydroxide supplemented with 0.1 mM biotin and then washed an additional three times with Wash Buffer and 0.1 mM biotin. The beads were then re-suspended in Hybridization Buffer (20mM Tris pH 8.0, 1 M NaCl, 0.1% Tween-20) supplemented with 0.1 mM biotin.

[00102] The pooled beads were split into ten reactions to which one of ten partially complementary oligonucleotides (Table 3) each containing a specific second barcode was added at a final concentration of 5 μΜ. The second barcode-containing oligonucleotides were allowed to hybridize to the beads at room temperature overnight on a rotisserie.

Table 3

[00103] Table 3 describes oligonucleotide sequences used to generate the second set of barcoded beads (SBC) for combinatorial synthesis in Experiment 1.

[00104] The beads were then washed five times in Wash Buffer supplemented with 0.1 mM biotin and then re-suspended in a reaction mixture with final concentrations of 0.5 mM dNTPs, lx NEB Buffer 2 (New England BioLabs), and 0.1 mM biotin. We included biotin in the wash and storage buffers in order to saturate any remaining streptavidin sites on the beads so that, in the even that a barcoded capture primer dissociates form a beads, it cannot re-associate with a different bead. The reactions were cooled to 16°C on a thermocycler and Klenow Fragment (exo-) (New England BioLabs) was added at a final concentration of 0.25 U^L. The reaction was incubated for one hour at 16°C with mixing every 10 minutes with a pipette followed by heat inactivation at 75°C for 20 minutes.

[00105] The ten reaction mixtures were then quenched and the hybridized strand was denatured by addition of sodium hydroxide to a final concentration of 125 mM. The reaction mixtures were then washed five times in 125 mM sodium hydroxide with 0.1 mM biotin, pooled, and then further washed three times with Wash Buffer supplemented with 0.1 mM biotin.

Procedure for Single Cell RNA-Seq Experiment 1

[00106] Prior to the experiment, each lane of the device was flushed with 0.1% Tween-20 solution and incubated for several hours to hydrate the micro wells, which were subsequently washed with 2 mL of phosphate-buffered saline (PBS). Cell suspensions were counted using Countess automated cell counter (Life Technologies). A suspension of cells in PBS mixed with Calcein AM (live stain) dye was flowed in to each lane and incubated for ~5 mins, so that the cells load in to the microwells under gravity. After thoroughly washing out the excess cells with PBS, a suspension of barcoded capture beads that had been pre-counted by microscopy was introduced in PBS and allowed to load under gravity for ~5 mins. In an aspect, -3,000 cells to each lane of the device was introduced. It is also noted that only 25% of the lower surface of each channel contains a microwell array, and so by expanding this area, the number of cells captured without incurring increased reagent costs for on-chip library generation were significantly increased (for example, as long as the size of the barcode pool was pooled). Excess beads were washed out thoroughly with PBS and the flow cell was incubated on ice. 20μί 0.08% TritonX-100 (Sigma) supplemented with SUPERaseIN in PBS was flowed under ice-cold conditions immediately followed by fluorinert oil (Sigma) to seal the device. After two cycles of freeze-thaw at -80°C to enhance cell lysis, the device was incubated at room temperature for 60 mins for mRNA capture (FIG. 2A).

[00107] Two of the lanes contained pure U87 and MCFlOa cells, respectively, and other lanes were loaded with a mixture of both the cell types. All lanes were imaged twice, first with blue laser (λ =473 nm, Dragon Lasers) for imaging the cells and secondly with a red laser (λ =637 nm, Obis, Coherent) for imaging the beads labeled with AlexaFluor 647 tagged streptavidin. We used the two-color images to determine number of bead-cell pairs in the array. After an hour of incubation for mRNA capture, all the lanes were unsealed by rapid washing of the oil with 20mM Tris, containing 1% TritonX-100 and SUPERaselN, followed by Wash Buffer supplemented with

SUPERaselN. After this point the microwells stay open and subsequent enzymatic steps occur simultaneously in separate lanes of the open device.

[00108] The single cell library preparation protocol is adopted from the recently reported CEL-Seq protocol with few modifications as described below. The mRNA captured on the beads was reverse transcribed using ProtoScript II Reverse Transcriptase (New England Biolabs) for 2 hours at 42°C in lx ProtoScript Reverse Transcriptase buffer, supplemented with lOmM DTT, 0.5mM dNTPs, 0.1% Tween-20 and

SUPERaselN. The reaction mixture was washed out with Wash Buffer. The second strand synthesis was carried out using reagents from the MessageAmp II aRNA amplification kit (Ambion), where a mixture of DNA polymerase and RNaseH in second strand buffer was used along with dNTPs by incubating the device at 16°C for 2 hours. After flushing out the second strand reaction mixture with Wash Buffer, an in vitro transcription mixture from the MessageAmp II kit containing four nucleotides and T7 RNA polymerase enzyme mix in T7 buffer was introduced to all lanes and incubated for 13 hours at 37°C (FIGS. 5A - 5C). The reaction linearly amplified our cDNA, eluting pools of barcoded aRNA into the flow channels of the device which was then removed from each lane using a pipette and purified separately using RNA Clean & Concentrator columns (Zymo) and eluted into five separate tubes. The aRNA from the 5 lanes was reverse transcribed separately using random hexamers tagged with five different barcodes and 8-base UMIs to differentiate cDNA for all five lanes and part of an Illumina sequencing adapter. The aRNA along with the hexamer primers was heated to 70°C for 2 mins and immediately placed on ice for 5 mins. The reverse transcription mix containing PrimeScript Reverse Transcriptase (Clontech-Takara), 0.5mM dNTPs, lOmM DTT, lx PrimeScript buffer supplemented with SUPERaseIN was added and incubated at 25°C for 10 mins followed by 2 hour incubation at 42°C. The RNA-cDNA hybrid product was purified twice using 0.65x ratio of Agencourt Ampure beads (Beckman Coulter) and the purified cDNA from all the lanes were pooled together for PCR. Phusion High Fidelity DNA polymerase (New England Biolabs) was used for amplifying the cDNA using RP1 and RPI Illumina primers in lx PhusionHF buffer supplemented with dNTPs. The PCR product was purified on a 1.5% agarose gel which was stained with SybrGold (Life Technologies) before being cut between 400-800 bp. The library was extracted from the gel using Gel Extraction kit (Qiagen), and further purified and concentrated using a 0.65x ratio of the AMpure beads (Beckman Coulter). The final library was quantified using a Qubit (Life Technologies) and Bioanalyzer (Agilent) and sequenced on NextSeq 500 desktop sequencer (Illumina). -240 million paired-end reads with a 26-base first read and a 66-base second read were obtained.

Procedure for Single Cell RNA-Seq Experiment 2

[00109] Experiment 2 was identical to Experiment 1 with a few exceptions. First, the two cell types under study were U87 human glioma cells and WI-38 human fibroblast cells (a diploid, limited-passage, non-cancer cell line). Second, reagents from the HiScribe In Vitro Transcription kit (New England BioLabs) were substituted for the MessageAmp II kit for the IVT portion of the protocol. Third, some of the oligonucleotides used were different from in Experiment 1 and are tabulated as detailed in Tables 4-6.

[00110] Table 4 describes a group of preferred oligonucleotides used for barcoding and library preparation for Experiment 2.

Table 4 RT1 (Reverse Transcription Primer for Lane 1) GCCTTGGCACCCGAGAATTCCANNNNNNNNCGTCATN

NNNNN

RT2 (Reverse Transcription Primer for Lane 2) GCCTTGGCACCCGAGAATTCCANNNNNNNNTACCCAN

NNNNN

RT3 (Reverse Transcription Primer for Lane 3) GCCTTGGCACCCGAGAATTCCANNNNNNNNGCCATTN

NNNNN

RT4 (Reverse Transcription Primer for Lane 4) GCCTTGGCACCCGAGAATTCCANNNNNNNNGAGTACN

NNNNN

RT5 (Reverse Transcription Primer for Lane 5) GCCTTGGCACCCGAGAATTCCANNNNNNNNAGAGTCN

NNNNN

RP1 (PCR Primer 1) AATGATACGGCGACCACCGAGATCTACACGTTCAGAG

TTCTACAGTCCGA

RPI2 (PCR Primer 2) CAAGCAGAAGACGGCATACGAGATACATCGGTGACTG

GAGTTCCTTGGCACCCGAGAATTCCA

[00111] Table 5 describes oligonucleotide sequences used to generate the first set of barcoded beads (FBC) for_combinatorial synthesis in Experiment 2.

Table 5

FBC_01igo24 CAGGTCGTTGGTGTTGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo25 CAGGTCCCTTCAGACGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo26 CAGGTCCCTCTTGGTGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo27 CAGGTCGGGAAAGTTGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo28 CAGGTCAGCCAGAGTGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo29 CAGGTCTCGCATCTGGATCGTCGGACTGTAGAACTCTGAAC

FBC_Oligo30 CAGGTCGATACGGCAGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo31 CAGGTCTCGGCCAAAGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo32 CAGGTCAGATTTCGCGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo33 CAGGTCGACCCTCAAGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo34 CAGGTCAGTCCACTCGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo35 CAGGTCCAAACGATCGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo36 CAGGTCGCCTAATAGGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo37 CAGGTCGGCTACATCGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo38 CAGGTCTATGAGCAGGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo39 CAGGTCGGTAGTAACGATCGTCGGACTGTAGAACTCTGAAC

FBC_Oligo40 CAGGTCCGCGTATATGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo41 CAGGTCTACTGGAGCGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo42 CAGGTCAGGGAATCAGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo43 CAGGTCATCCGAGATGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo44 CAGGTCTCCCAAGCAGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo45 CAGGTCGAGCCGTTTGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo46 CAGGTCTGCTCTTACGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo47 CAGGTCACGACTACCGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo48 CAGGTCCAAGCAGCTGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo49 CAGGTCGTATTCGCGATCGTCGGACTGTAGAACTCTGAAC

FBC_Oligo50 CAGGTCGCTCTGAAGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo51 CAGGTCACGTAGTGGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo52 CAGGTCATTGGGTCGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo53 CAGGTCAACAGCACGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo54 CAGGTCTCAGAGACGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo55 CAGGTCGTGTGCTAGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo56 CAGGTCGCAGTTGAGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo57 CAGGTCTTAACGGGGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo58 CAGGTCGCTCGATTGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo59 CAGGTCACACCTGTGATCGTCGGACTGTAGAACTCTGAAC

FBC_Oligo60 CAGGTCAGACGGTTGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo61 CAGGTCGCAAACCAGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo62 CAGGTCGAGTATGGGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo63 CAGGTCGGTCTTTCGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo64 CAGGTCCATCTGCTGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo65 CAGGTCTTCGCAAGGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo66 CAGGTCTTGTGACGGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo67 CAGGTCTGCATGACGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo68 CAGGTCCAACGTGAGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo69 CAGGTCTAGGCTTCGATCGTCGGACTGTAGAACTCTGAAC

FBC_Oligo70 CAGGTCTGGTAGGAGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo71 CAGGTCTGCAGCTTGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo72 CAGGTCCTGTACCTGATCGTCGGACTGTAGAACTCTGAAC FBC_01igo73 CAGGTCCGCAATGAGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo74 CAGGTCGATCCAAGGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo75 CAGGTCCACTTACGGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo76 CAGGTCAACTAGGCGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo77 CAGGTCACTAGCGTGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo78 CAGGTCCGTTCGTTGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo79 CAGGTCAGTCACGAGATCGTCGGACTGTAGAACTCTGAAC

FBC_Oligo80 CAGGTCCCTGTAACGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo81 CAGGTCGTCCTCTTGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo82 CAGGTCCAGCGAATGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo83 CAGGTCATGGTTGGGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo84 CAGGTCGAGGTTCTGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo85 CAGGTCTACCTCGAGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo86 CAGGTCTTCTGTGCGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo87 CAGGTCGACAACTGGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo88 CAGGTCCGACAACAGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo89 CAGGTCTCGATACCGATCGTCGGACTGTAGAACTCTGAAC

FBC_Oligo90 CAGGTCCCATACTCGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo91 CAGGTCATTCGCAGGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo92 CAGGTCACCATAGGGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo93 CAGGTCCGATCAAGGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo94 CAGGTCACCTTGCTGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo95 CAGGTCGACTCAGTGATCGTCGGACTGTAGAACTCTGAAC

FBC_01igo96 CAGGTCGTCAATCCGATCGTCGGACTGTAGAACTCTGAAC

[00112] Table 6 provides for oligonucleotide sequences used to generate the second set of barcoded beads (SBC) for combinatorial synthesis in Experiment 2.

Table 6

AAAAAAAAAAAAAAAAAAAAAAAAAG T G C AG T C AG GT C AAAAAAAAAG AT C G

SBC_01igo 9 TCGGACTGTAGAACTC

AAAAAAAAAAAAAAAAAAAAAAAAAAG G TAG AC AG GT C AAAAAAAAAG AT C G

SBC_01igo l O TCGGACTGTAGAACTC

Analysis of Single Cell RNA-Seq Data

[00113] Read 1 of the single cell RNA-Seq data contains a cell-identifying barcode sequence followed by poly(dT), and read 2 contains a 8-base UMI followed by a 6-base lane-identifying barcode and a transcript sequence. The reads are first demultiplex based on the lane-identifying barcode while recording the corresponding UMI using a custom

Python script. We then map the remainder of read 2 to the human genome and

transcriptome (hgl9, Ensembl annotation from Illumina iGenomes) using the STAR

aligner. Mapped reads for each lane are then demultiplexed based on the cell-identifying barcodes in read 1 and assigned to a gene using HTSeq. Both the lane- and cell- identifying barcodes were allowed to have a single-base mismatch during demultiplexing.

[00114] The set of reads that uniquely mapped to the transcriptome were collected and assigned an address comprised of its cell-identifying barcode, gene, UMI, and

mapping position. In addition, the reads that mapped to both the genome and

transcriptome were kept, but that mapped to only one position on the transcriptome and mapped to that position with the appropriate strand-specificity. The reads to identify

unique molecules were filtered. Reads with identical addresses were collapsed to a single molecule. In addition, reads with identical cell-identifying barcodes, genes, mapping

positions, and with UMIs having a Hamming distance less than or equal to two were

collapsed to a single molecule. All reads considered identical molecules by the above definition (UMIs with a Hamming distance less than or equal to two and mapping

position within six bases) were removed but that also occurred with different cell- identifying barcodes within the same lane. This approach likely underestimates of the true number of molecules associated with each cell and gene and results in some loss of gene detection. However, it also removes molecules that may become spuriously

associated with the incorrect cell via PCR recombination, as observed and similarly

filtered in previous studies that used very similar library construction protocols. [00115] To identify barcodes that correspond to actual individual cells in our device in Experiment 1, the observed cell-identifying barcodes were filteredby progressively downsampling the corresponding gene profiles to the same number of total reads and assessing the number of unique molecules detected from each cell-identifying barcode. After excluding cell-identifying barcodes having zero associated molecules, it was found that the distribution of associated unique molecules to be bimodal, with one small subpopulation having nearly as many unique molecules as reads at low read totals. It was found the size of this subpopulation to be in excellent agreement with our device imaging data. These 598 profiles were taken to represent the actual individual cells captured in the device with a barcoded bead. We used the same approach to assess the cell-identifying barcodes in Experiment 2.

[00116] The 396 single-cell profiles were kept with the highest coverage in the data set (all five lanes represented). The U87 and MCFlOa single cell profiles were compared to bulk RNA-Seq profiles of U87 and MCFlOa cells. Bulk RNA-Seq library from ~10 ⁷ U87 cells were prepared using the TruSeq RNA-Seq library preparation kit (Illumina) and sequenced the library to a depth of ~30M, 100-base single-end reads on an Illumina HiSeq 2500. Publically available bulk RNA-Seq profile of MCFlOa cells was obtained from the Gene Expression Omnibus (entry GSE45258). Reads were mapped to the transcriptome as described above and expression values (FPKM) were computed using Cufflinks. Pearson correlation coefficients between single cell and bulk profiles were computed between log-transformed single-cell expression profiles (unique molecules per million reads plus one) and log-transformed bulk values (FPKM plus one). Single cell median profiles were generated from different numbers of randomly selected single cell profiles and repeated this random sampling ten times without replacement for each data point in FIGS. 7A and 7B. For each Pearson correlation calculation, only genes with log-transformed single cell median or bulk values greater than 0.5 were included.

[00117] Differential expression analysis was conducted by comparing each detected gene in the two cell type-exclusive lanes using Wilcoxin' s rank-sum test. Genes with p<0.05 were used for clustering analysis. Regardless of differential expression +/- (1-p) was used (which is positive for expression biased in one cell type and negative for expression biased towards a second cell type) for each gene as input to iPAGE, a mutual information-based algorithm that can associate gene ontologies with genes based on an assigned numerical value[35]. A matrix of pairwise Spearman correlation coefficients was generated based on unique molecules detected across 396 single cell profiles in Experiment 1 (247 profiles in Experiment 2) using only the differentially expressed genes. The data was then clustered with the MATLAB implementation t-SNE using the correlation matrix as input. The single cell profiles were color coated in the t-SNE clusters using a simple classifier score given by the log-ratio of the number of cell type- specific genes for each of the two cell types in a given cell with an above-average rank in expression level (FIG. 8C).

[00118] Table 7 describes estimated costs per reagents and associated costs in an aspect of the disclosure.

Table 7