The problem. After clustering, someone has to say what each cluster is — which cell type. Done manually, that means hunting marker genes cluster by cluster: slow, subjective, and inconsistent between analysts. As datasets grew to hundreds of clusters across many samples, hand-annotation stopped scaling.

The idea. SingleR automates it by correlation. Given reference transcriptomes of known cell types, it scores each single cell (or cluster) against every reference profile and assigns the best-matching label, with a fine-tuning step to resolve close calls. The annotation becomes reproducible and reference-driven rather than a judgment call — and the paper demonstrates it by uncovering a transitional macrophage state in fibrotic lung.

Why it matters. Automated annotation is a standard step, and SingleR’s reference-correlation approach connects to the recurring “transfer labels from reference to query” theme — scANVI (day 4) and Seurat anchors (day 10) do versions of the same thing. It also foreshadows the foundation-model framing (Geneformer, scGPT) where a pretrained reference does the labelling. Different mechanisms, one goal: stop annotating by hand.

Verdict. Foundational for automated cell-type annotation and widely used; its labels are only as good as the reference, so reference choice matters. Read it as the correlation-based answer to “what is this cluster,” and part of the label-transfer family running through the reading list.