Single-cell omics

Single-cell omics maps the genome, transcriptome or proteome of each individual cell rather than the bulk average of a tissue, turning a “smoothie” of mixed cells into a catalogue of distinct cell types. It is delivered overwhelmingly as a contracted service: a biotech or clinical research group sends a fresh tissue biopsy to a platform or CRO, which runs droplet-based single-cell RNA sequencing (scRNA-seq), returns a cell-by-cell gene-expression matrix, and analyses it with machine-learning pipelines. The industry-standard capture uses microfluidic gel-beads-in-emulsion (GEM) on the 10x Genomics Chromium platform, where each cell is lysed inside a droplet and tagged with a 16-nucleotide cell barcode and a 12-nucleotide unique molecular identifier (UMI); CITE-seq extends this to surface proteins via DNA-barcoded antibodies, and scATAC-seq reads chromatin accessibility per cell. The output powers oncology (rare drug-resistant tumour subclones), immunology (T-cell exhaustion and checkpoint-inhibitor response) and the international Human Cell Atlas. Deep sequencing needs at least 50,000 reads per cell, so the service bundles microfluidics, Illumina-grade sequencing and AI analysis into one CRO workflow that biopharma buys by the project.

The key directions of single-cell omics are:

1. scRNA-seq capture platforms: microfluidic droplet barcoding (10x Chromium, BD Rhapsody, Singleron) that isolates and barcodes tens of thousands of cells per run for transcriptome-wide expression.
2. Multi-omics per cell (CITE-seq, scATAC-seq): simultaneous readout of surface proteins (CITE-seq) or chromatin accessibility (scATAC-seq) alongside RNA, giving full phenotype resolution.
3. Spatial and trajectory methods: spatial transcriptomics preserves tissue location, and pseudotime algorithms (Monocle, scVelo) order static cells along a developmental trajectory.
4. AI analysis services: pipelines (Cell Ranger, Seurat, Scanpy) that demultiplex reads, cluster cells by type on UMAP maps and surface unique oncomarkers for the client.

Sectoral value chain#

[Viable cell isolation] ──> [Microfluidic GEM barcoding] ──> [NGS library assembly]
                                          │
                               (deep sequencing)
                                          │
                                          ▼
[Therapeutic-target discovery] <─── [AI clustering / Seurat] <─── [Cell Ranger pipeline]

Value chain levels#

Cross-cutting technologies of the sector:

• Microfluidic GEM barcoding: precise water-in-oil droplet generation where each droplet holds one cell and one gel bead coated with millions of primers carrying a shared cell barcode and a unique molecular identifier.
• Unique molecular identifiers (UMIs): short random tags on each mRNA that let the pipeline count original RNA molecules absolutely, cancelling out PCR amplification bias.
• Pseudotime trajectory inference: computational ordering (Monocle, scVelo) of static single cells along a virtual developmental or activation trajectory from smooth changes in their expression profiles.

The single-cell-omics service stack joins a microfluidic capture engine, multi-omics chemistry and an AI analysis layer into one CRO workflow.

1. Microfluidic GEM capture and UMI barcoding:
   • Droplet platforms (10x Chromium, BD Rhapsody, Singleron) co-encapsulate one cell and one barcoded gel bead per droplet; the bead carries a 16-nt cell barcode and a 12-nt UMI on every primer.
   • UMIs let the pipeline count original RNA molecules absolutely, removing PCR amplification bias so expression values are quantitative across cells.
2. Multi-omics per cell (CITE-seq, scATAC-seq):
   • CITE-seq labels surface proteins with DNA-barcoded antibodies that are read alongside the mRNA, giving full phenotype in the same cell, while scATAC-seq maps open-chromatin regions.
   • Mission Bio’s Tapestri extends per-cell readout to somatic DNA mutations plus proteins, used for minimal-residual-disease monitoring in leukaemia.
3. AI analysis and trajectory inference:
   • The Cell Ranger pipeline demultiplexes FASTQ reads, aligns them to the genome and builds the gene-by-cell matrix that downstream tools consume.
   • Seurat and Scanpy reduce dimensionality (UMAP), cluster cells into types and infer pseudotime trajectories (Monocle, scVelo), delivering the client a UMAP cluster map and a list of unique oncomarkers.