One Tissue, Many Cells

The everyday analogy

Suppose a hospital asks whether patients are waiting too long. One number for the whole hospital hides the difference between emergency, radiology, pharmacy, and reception. Measuring each department separately reveals where the delay occurs.

Bulk RNA-seq is the hospital-wide average. Single-cell RNA-seq is closer to measuring each room separately. It can reveal a rare population, a changing cell state, or a different mixture of cells that the average hides.

What RNA tells us

DNA is a long-term instruction store. A gene is a named region that can contribute to a biological product. When a gene is active, the cell can make RNA copies called transcripts. Messenger RNA is not a direct measurement of protein, behavior, or disease, but its abundance is evidence about what a cell was doing when it was captured.

Single-cell RNA-seq asks: which RNA molecules were observed in each captured cell or nucleus?

The usual result is a matrix:

Rows are cells, columns are genes, and entries are observed molecule counts. Most entries are zero, so the matrix is sparse.

Questions it can answer

Single-cell analysis is useful when the variation among cells matters:

• Which cell populations are present in a tumor?
• Does treatment change a cell type’s abundance or state?
• Which immune population expresses an inflammatory program?
• What intermediate states appear during development?
• Which cells respond to infection?

It is less useful when the scientific question concerns only a well-purified, uniform population or when sample replication is too weak to support the comparison.

Four viewpoints

The biologist asks whether the populations and markers make biological sense. The computational researcher asks whether processing choices created the pattern. The programmer asks whether data structures and algorithms scale. The clinician asks whether the finding was replicated and whether it changes a validated decision.

All four viewpoints are needed. A technically correct cluster can still be a doublet. A plausible marker can still be caused by batch. A statistically significant gene can still be clinically irrelevant.

Checkpoint

Before continuing, explain this sentence in your own words:

Single-cell RNA-seq measures noisy evidence about cell state, not a complete inventory of everything a cell contains or does.