The problem. Single-cell tooling moves so fast that “best practice” is a moving target, and the 2019 tutorials — good as they were — no longer cover multimodal data, the deep-generative methods, or the pitfalls the field has since learned. Newcomers need one current, opinionated, end-to-end reference.

The idea. This is a comprehensive review spanning the whole workflow — QC, normalization, integration, clustering, annotation, trajectory, differential and compositional analysis — and reaching across modalities (RNA, ATAC, protein, spatial). Crucially it’s tied to the living sc-best-practices.org book, so the recommendations come with runnable code and get updated as methods change, rather than freezing at publication.

Why it matters. This is the natural capstone to the spatial and single-cell fortnight: everything I read this week — scVI, scANVI, the deconvolution and segmentation methods — has a place in this map, and the review is where the judgment about when to use each lives. For a facility, a shared, current best-practice reference is worth more than any single tool; it’s the standard you point collaborators to.

Verdict. The most useful “one thing to bookmark” of the batch, and the maintained book is the real deliverable. Its only ceiling is the field’s own churn — even a living document lags the newest methods. Read it as the framework, and treat the website as the working manual. This closes out §6; tomorrow the reading turns to foundations.