Single-cell RNA-seq
Cells, barcoded gel beads and oil meet in a microfluidic chip, forming droplets in which each cell's poly(A) RNA is reverse transcribed with a cell barcode and UMI. Thousands to tens of thousands of cells per channel, counting the 3′ (or 5′) end of transcripts.
Our ruleSimulations prepare you for the bench but don't replace hands-on practice.
Droplet-based (10x)
Cells, barcoded gel beads and oil meet in a microfluidic chip, forming droplets in which each cell's poly(A) RNA is reverse transcribed with a cell barcode and UMI. Thousands to tens of thousands of cells per channel, counting the 3′ (or 5′) end of transcripts.
How it differs from the standard method
- Load by the target cell number; multiplets scale with cells recovered.
- About 60% of loaded cells are typically recovered.
- Typically ~20,000–50,000 reads per cell; 1,000–5,000 genes detected per cell, depending on cell type.
- Empty droplets measure the ambient RNA profile.
1One barcode per cell
Bulk RNA-seq averages thousands of cells, so a change in one rare population, or a shift in which cells are present, is diluted or misread as a change in gene regulation. Single-cell RNA-seq labels every captured mRNA with a barcode that identifies its cell before the material is pooled.
Thousands of cells, 3′ counts
Cells and barcoded beads meet in nanolitre droplets. Cheap per cell and high throughput; counts the 3′ end of each transcript with UMIs.
Hundreds of cells, full length
Index-sorted cells in wells, full-length cDNA by template switching. More genes per cell and isoform information, at higher cost per cell.
RNA plus surface protein
Antibodies carrying DNA tags are bound to the cells before capture; the tags are counted alongside the RNA.
2Numbers that decide the experiment
A droplet is loaded by chance, so the number of cells in it follows a Poisson distribution. Loading more cells raises the chance that a droplet holds two, and the multiplet rate rises roughly in step with the number of cells recovered (about 0.8% per 1,000 in common kits).
3What the clusters mean
Cells are grouped by similarity of their expression, usually drawn on a two-dimensional embedding such as UMAP. A cluster is a hypothesis about a cell type or state, not a fact: doublets form bridges or small clusters between real types, stressed cells form a parallel cluster, and ambient RNA makes every cell look a little like the most abundant one.
Key terms
Practise these as flashcards →A sequence shared by every primer on one bead (or in one well) that labels all transcripts from that cell.
A random sequence on each primer that marks each captured molecule, so PCR copies are counted once.
A barcode that holds two or more cells; heterotypic doublets mimic intermediate states, homotypic ones are invisible.
Free RNA in the suspension, mostly from lysed cells, captured in every droplet along with the cell.
The share of reads that come from a molecule already seen; high saturation means more reads add few new UMIs.
UMIs per barcode against rank; cells form a plateau, empty droplets a lower shoulder, and a sharp knee between them indicates a clean run.
The reverse transcriptase adds a few cytosines at the end of the cDNA, where an oligo anneals so full-length cDNA can be amplified (Smart-seq).
The DNA barcode on a CITE-seq antibody, counted like a transcript to measure surface protein per cell.
When to use it
- Finding which cell types and states are present in a tissue, including rare ones.
- Deciding whether a bulk expression change comes from regulation within cells or from a shift in cell-type proportions.
- Following differentiation trajectories or responses that differ between cell types.
- Adding surface-protein measurements (CITE-seq) when states differ more in protein than in RNA, or to link to flow cytometry gates.
Compared with alternatives
Bulk RNA-seq is cheaper, deeper per gene and better powered for differential expression when the population is homogeneous or the question is about the average.
Measures a panel of proteins on millions of cells quickly and can sort them; prefer it when markers are known and numbers matter.
Keeps the tissue position that dissociation destroys, at lower resolution or for fewer genes.
Long-read single-cell libraries resolve full-length isoforms per cell.
Sub-techniques
Choosing one here switches the whole page, including the protocol and the lab, to it.
Cells, barcoded gel beads and oil meet in a microfluidic chip, forming droplets in which each cell's poly(A) RNA is reverse transcribed with a cell barcode and UMI. Thousands to tens of thousands of cells per channel, counting the 3′ (or 5′) end of transcripts.
- Load by the target cell number; multiplets scale with cells recovered.
- About 60% of loaded cells are typically recovered.
- Typically ~20,000–50,000 reads per cell; 1,000–5,000 genes detected per cell, depending on cell type.
- Empty droplets measure the ambient RNA profile.