Fig 1.
Screening genetic interaction libraries by single-cell sequencing with droplet microfluidics.
(a) Genetic libraries can be genomically encoded or introduced through episomal DNA like plasmids. Interaction libraries are created by combining two genetic libraries. Some of the most common types of interaction libraries are noted. (b)Libraries are screen by microfluidic encapsulation and single-cell linkage PCR (scLPCR) inside picoliter droplets. Confining cells inside of droplets allows PCR to link cellular DNA without crossover contamination of DNA from other cells. The PCR products are sequenced using paired-end chemistry on an Illumina platform to decode linkage products.
Fig 2.
Droplet based single-cell sequencing preserves genomic structure and population membership.
(a) KEIO collection strains of E. coli used to test linkage PCR: a barcode has been inserted into the genome at defined loci, creating gene knockouts of ynaA and ybfH in strains ECK1365 and ECK0679, respectively. (b) Linkage PCR to fuse the sequence from both genomic loci in the two strains yields a mixture of four products in bulk (left), two of which reflect the true genomic organization and two that reflect spurious mixed cell products. However, single-cell linkage PCR (scLPCR) only yields the two products that reflect true genomic organization. (c) Deep sequencing of products from bulk linkage PCR or scLPCR showing percent of reads that reflect true genomic organization (in black) or spurious mixed cell products (in red), indicating the scLPCR on a culture of mixed cell types recovers reports on the genomic variation within the population. (d) The fraction of the population determined by sequencing depth (red dots) when one KEIO strain is spiked into a culture of the other strain at defined dilutions shown on the x-axis. The expected results are shown as a dashed line.
Fig 3.
Screening complex libraries with droplet sc-Seq.
(a) Library of E. coli containing 64 strains, each containing a pair of known barcodes (denoted X and Y) located on separate plasmids. (b) The accuracy of barcode X and Y pairing from NGS (as percent of sequencing reads that report a correct X/Y pair) is the same when using linkage PCR on isolated strains (well plate LPCR) or when using single cell linkage PCR in droplets (Droplet scLPCR), while linkage PCR from all strains in bulk (Bulk LPCR) yields mostly random X/Y pairs. (d) The amount of mutual information between specific X/Y barcode pairs in the NGS data shows strong correlations along the diagonal, representing true X/Y pairs. In the same data for libraries from linkage PCR in bulk there is no correlation between represented barcodes.
Fig 4.
Screening a combinatorial library of amino acid auxotrophy with sc-Seq.
(a) The membership (by strain) of heterogeneous cultures is tracked by droplet scLPCR for cultures grown in rich media (RM) or minimal media (MM) at 0,4,8, or 16 doublings after inoculation. (b) The fractional fold change in minimal media vs. rich media over 16 doublings shows that auxotrophic strains with no complement drop out of the population (ΔtyrA, black line) while prototrophic experience no selection and take over (ΔynaA, red line). (c) Droplet scLPCR shows the culture composition by strain and plasmid and unmasks the mechanism of complementation, whereby auxotrophic strains persist in the culture through selective outgrowth of only those strains that harbor the corresponding complementary gene (color corresponds to fraction of sequencing reads within each strain that are specific for the corresponding complement gene).