Updated: Aug 27, 2021
Single-cell sequencing refers to the sequencing of individual cells to profile its genomic, transcriptomic, and other multi-omic information, in comparison to the traditional method of acquiring these multi-omic profiles from the average of cell populations. With the advancement of sequencing technologies and lower sequencing cost, single-cell sequencing technologies are more widely applied in various fields, to better understand cellular heterogeneity and confer the potential of detecting novel subpopulations.
In addition, recent technologies have provided the opportunities for understanding individual cells in a holistic approach by integrating information from multiple modalities, delineating a complete cellular map. We can now simultaneously detect the gene expression data with their spatial information at almost single-cell resolution, for example, with the 10X Visium technology. We can pinpoint the location of cells of interest on the histological slide, based on the clustering annotation (Fig. 1). This allows the identification of spatiotemporal gene expression patterns that may be interesting/important to any study.
Fig. 1 Visium data generated from the anterior section of sagittal mouse brain slices (originally reported in the Seurat spatial vignette tutorial from Satija Lab).
There are also sequencing platforms (e.g. Mission Bio's Tapestri platform), which connect genotypes to phenotypes, by sequencing both genome and proteome of individual cells, unraveling the relationship between genetic variants and their functional impacts. Furthermore, we can integrate single-cell data from different experiments which share common features, like similar cell types, to complement the information lacking in a single modality experiment, using batch-effect correction methods to identify a shared subspace or equivalent cells for these groups (Fig 2). Overall, we can piece together these data modalities of single cell and form a linkage between them with the integrated study.
Fig. 2 Multiple data sets are computationally integrated, using canonical correlation analysis (CCA) or mutual nearest neighbours (MNNs), to remove dataset specific variations, allowing direct comparisons between them (adapted from Stuart, T. & Satija, R., 2019).
Dr Loh Jui Wan
Bioinformatics Lead, Cancer Discovery Hub
National Cancer Centre Singapore