Single cell genomics: advances and future perspectives

Abstract

Advances in whole-genome and whole-transcriptome amplification have permitted the sequencing of the minute amounts of DNA and RNA present in a single cell, offering a window into the extent and nature of genomic and transcriptomic heterogeneity which occurs in both normal development and disease. Single-cell approaches stand poised to revolutionise our capacity to understand the scale of genomic, epigenomic, and transcriptomic diversity that occurs during the lifetime of an individual organism. Here, we review the major technological and biological breakthroughs achieved, describe the remaining challenges to overcome, and provide a glimpse into the promise of recent and future developments.

Figure 1. Detection of various classes of genetic variation using single-cell WGA-NGS approaches.

A) The most prominent methods for (i–ii) isolating individual cells (including (i) creation of single-cell suspensions—usually by enzymatic tissue disaggregation—and subsequent cell isolation through manual micro-pipetting , , , , fluorescence-activated cell sorting , or microfluidics devices , , , and (ii) laser capture microdissection [109], [110]) as well as (iii) isolating single nuclei , , , are indicated, accompanied with particular advantages and disadvantages. A comprehensive review of single-cell isolation methods is presented by Shapiro et al. . B–D) Subsequently, the cell is lysed and its genome amplified. A standard sequencing library can be prepared from the WGA product for paired-end (or single-end) sequencing. The resulting (short) sequence reads of the cell are mapped against a reference genome for variant discovery (Ei–Eiii). In all steps (Ei–Eiii towards F), various confounding factors resulting from the WGA process have to be considered in the analysis (indicated in red boxes). Ei–F) Structural variants can be detected by analysing read-pairs which map discordantly to the reference genome, or by discovering split reads crossing a rearrangement. However, WGA can create various chimeric DNA molecules that resemble structural variants following paired-end sequence analysis of the WGA-product. Eii–F) Copy number variants are called by “binning” reads that map to particular regions of the genome. By comparing the read count per bin to the counts obtained in a reference sample , or an average read count per bin , a copy number profile can be calculated. However, single-cell copy number profiles can be distorted by ADO, PA, and %GC-bias during the WGA process. Eiii–F) Single nucleotide variants (SNVs) can be detected in sequenced single-cell WGA products by aligning the reads with a reference genome. However, three cells carrying the same SNV are required to confidently call the variant.

Figure 2. Overview of single-cell WGA approaches.

A) Multiple displacement amplification (MDA) initiates with random priming of denatured single-cell DNA template, followed by a 30°C isothermal amplification using a DNA-polymerase with strand-displacement activity, typically phi29 . When the 3′-end of a newly synthesized fragment reaches the 5′-end of an adjoining primed string of nucleotides, it will displace the latter, liberating single-stranded DNA for new primer annealing and DNA-synthesis. B) Primer extension pre-amplification (PEP)-PCR , degenerate-oligonucleotide primed (DOP)-PCR , and linker-adaptor (LA)-PCR use PCR for WGA. C) WGA methods like PicoPlex , and MALBAC use displacement pre-amplification to generate PCR-amplifiable fragments (abbreviated as DA-PCR methods here). Specifically, MALBAC initiates with multiple rounds of displacement pre-amplification using a primer that anneals randomly throughout the genome, but contains a specific sequence allowing full amplicons to form looped pre-amplification products of a cell's template DNA. This looping protects previously copied segments from further pre-amplification. Multiple rounds of the displacement pre-amplification reaction, interspersed by a denaturation step, increase the probability that random priming will occur across the genome. D) Classes of genetic variation reported in single cells following WGA and analysis. The proofreading capacity of φ29 improves sequence fidelity during WGA , , , . Furthermore, MDA amplifies the majority of a cell's genome and appears a preferred method for SNP genotyping , , or base mutation detection , , , , but ADO and PA occurs. Following MDA, single-cell copy number profiles can be distorted , —although improvements are emerging —and chimeric DNA amplification products are created , . In general, the (DA-)PCR-based WGA products more accurately preserve the copy number profile of the template genome – and can be used for SNP genotyping and base mutation detection .

Figure 3. Single-cell transcriptomics.

A) A single cell is thought to contain a few hundred thousand mRNA transcripts, present in a log-normal distribution of abundances, with as much as 85% speculated to be present between 1–100 copies. Current amplification methods are thought to reliably detect transcripts at >5–10 copies per cell. B) Single-cell transcriptomes reveal heterogeneity in gene (co)expression that bulk analysis would not permit. Six single cells are shown, with heterogeneous expression of three genes. C) Bulk analysis would detect uniform expression of all three genes. D–E) Single-cell analysis would reveal underlying heterogeneity but also indicate that two of these genes showed a pattern of co-expression. F) STRT-Seq , is initiated by reverse transcription using an oligo-dT adaptor-primer. At the 5′ cap of the transcript, non-template nucleotides are added by the reverse transcriptase, permitting hybridisation of a barcoded template-switching adaptor-primer. Following pooling of barcoded RT-products, PCR amplification is performed, after which the 5′-ends are captured and sequenced. G) CEL-Seq is initiated using a barcoded oligo-dT primer, which also contains a 5′ adaptor sequence and T7 RNA-polymerase priming site. Complimentary RNA is generated from the cDNA by T7 RNA polymerase. The cRNA is then fragmented and prepared for (3′-end) paired-end sequencing. H) The Tang/Surani method , , and improved derivatives , first generates, then 3′ polyadenylates, first strand cDNA. Priming with adaptor-conjugated oligo-dT generates double stranded cDNA, which is then amplified by PCR and sequenced. I) The SMARTer method , uses template-switching to generate a full-length transcript with adaptor sequences at both ends. These sequences are then used to prime PCR amplification of the transcriptome. The full-length cDNA is used as input for sequencing. J) Overview of the sequence coverage of a typical transcript which would be obtained by each of the currently available single-cell RNA-seq methods –.