Single-Cell RNA Sequencing: From Droplets to a Count Matrix
Droplet-based single-cell RNA sequencing aims to isolate individual cells and tags captured RNA-derived molecules with cell barcodes and molecular identifiers. It produces a sparse matrix of detected molecule counts. Follow the steps and artefacts that limit what those counts mean.
What a learner can do afterwards
- Trace captured RNA through lysis, reverse transcription, barcoding, amplification and sequencing to a detected molecule count for one gene and cell barcode
- Explain how unique molecular identifiers reduce PCR overcounting and why capture loss, tag errors and UMI collisions can still affect counts
- Name the artefacts the protocol introduces: doublets, empty droplets, ambient RNA and dropout
1 · Read
Cells sharing a genome can differ in RNA as well as protein. Here we measure RNA. A bead-based droplet protocol aims to load one cell and one barcoded bead per droplet, but many droplets have no cell and some have several. Lysis breaks cells open. Bead primers capture messenger RNA. During reverse transcription, RNA is copied into DNA, called cDNA, with a droplet barcode and a unique molecular identifier, or UMI. PCR amplifies the cDNA before sequencing.
Assign reads to genes, then group them by gene, cell barcode and UMI. Suppose one captured molecule gives 80 reads with matching assignments and tags. With no tag errors or UMI collisions, its group contributes one count. Do not merge different genes or droplets. UMIs reduce PCR overcounting, but tag errors and shared UMIs can affect estimates. In this genes-by-cells matrix, rows are genes, columns are retained cell barcodes, and entries estimate detected molecules.
UMIs cannot recover RNA that left no detected tagged copy. A zero may mean no RNA was present, or that present RNA was not captured, reverse-transcribed or sampled by sequencing. Such non-detection is often called dropout. A zero alone cannot show that a gene is off. RNA counts also do not directly measure protein abundance.
A doublet puts RNA from two cells under one droplet barcode. Mixed cell-type markers and unusually high RNA totals can suggest a doublet, but biological states can also show these patterns. Investigate doublets before naming an intermediate type. Removing flagged profiles does not prove that a remaining cluster has a distinct biological identity; that claim needs supporting biological evidence.
Ambient RNA is free RNA in the sample, often released by damaged cells. It can enter empty and cell-containing droplets; contamination need not be equal across droplets. RNA in empty droplets helps assess this background. Liver-associated RNA in empty droplets and unrelated cell types supports investigating ambient contamination. Transcription inside intact cells or doublets alone cannot explain RNA in droplets with no cells.
Barcodes identify droplets; UMIs reduce PCR overcounting. Counts describe detected RNA. Zeros, mixed profiles and background signals need further checks.
Lesson, questions, diagrams and flashcards written for this dataset.
Readings and flashcards written for this dataset.
More to read
What the count leaves out
A droplet barcode records a reaction compartment. The intended compartment holds one cell and a barcoded bead, but loading also produces empty droplets and droplets containing several cells.
After lysis, bead primers capture messenger RNA. Reverse transcription makes tagged DNA copies, and PCR amplifies those copies for sequencing. The analysis then uses gene assignments, cell barcodes and UMIs to group reads and estimate the number of detected molecules.
The matrix cannot list RNA that left no detected copy. Its entries are measurements of recovered material, not a complete inventory of the original cell. Loading errors also matter: one barcode can contain RNA from two cells.
Keep RNA and protein measurements separate
Single-cell RNA sequencing separates many RNA profiles that a pooled measurement would combine. Differences between profiles can reflect biology, technical sampling or both.
An RNA count does not directly report protein abundance. RNA must be translated, and RNA and protein can be lost at different rates. Nor does a snapshot of RNA abundance directly give the rate of transcription.
A zero is especially limited evidence. It may reflect no RNA in the cell, or present RNA missed during the experiment. Expression in neighbouring cells does not settle which explanation applies to this cell. Keep the measured quantity separate from the process you want to explain.
Check alternatives before naming a cell group
A profile containing markers from two cell types may come from a biological state or from two cells captured together. An unusually high RNA total adds a reason to investigate doublets, but neither clue is a definitive test.
Free RNA provides another explanation for unexpected markers. RNA released into the sample can enter droplets without cells and droplets containing unrelated cells. Empty droplets therefore help characterize ambient background. Contamination can differ across droplets.
Quality checks narrow the alternatives; they do not establish a cell's identity by themselves. After assessing doublets and ambient RNA, a proposed new type still needs supporting biological evidence. Report what was checked and which explanations remain possible.
2 · Watch
به انگلیسیTake it off screen
Where it sits
8 questions wait behind this lesson, each with its answer explained. Every answer feeds the sky: stars light as they are learned, and dim when it is time to come back.