The problem. Vendor pipelines get you from a Xenium run to a cell-by-gene matrix, but they don’t tell you whether to trust it, or what to check before analysis. This paper is the independent best-practice workflow — the QC discipline a facility needs and vendors under-specify.

The idea. It lays out quality assessment and an analysis workflow specific to Xenium in-situ data: evaluating transcript detection and background, assessing segmentation quality (the step where transcripts get assigned to cells — the dominant error source in imaging assays), filtering low-quality cells, and the normalization/clustering/annotation choices that follow, with the pitfalls called out at each stage. The through-line is that imaging-assay QC is different from sequencing-assay QC, because the failure modes (missegmentation, transcript misassignment, optical crowding) are different.

Why it matters. This is the paper I’d turn into a personal one-page workflow — “here’s how I’d analyze a Xenium dataset end to end, and here’s what I’d QC first.” That deliverable is precisely what a spatial core facility values: a reproducible, defensible pipeline with the checks made explicit. It also reinforces that segmentation is the make-or-break step, which is where the next two days of the reading list are headed.

Verdict. Being platform-specific (Xenium) is both its strength and its ceiling — the QC principles generalize, but the specifics won’t all transfer to MERSCOPE or CosMx, and chemistry updates will move details. Read it for the workflow skeleton and the QC mindset, then adapt. My plan: reproduce its checklist on a public Xenium dataset and post the worked version.