Your per-cell QC metrics, read sample by sample before you cut anything

Paste the metrics table — one row per cell, with the sample, the genes detected, the total UMI counts and the mitochondrial percentage. The browser computes every distribution, every MAD bound and the retention each threshold would produce, for free. The lane you pick does the judgement.

Your expression matrix never leaves your machine — only the metrics table you paste does. This is a worksheet, not a statistical conclusion: it documents the consequences of the thresholds you set and names what contradicts itself.

Both examples ship with a saved run, so you can see either lane end to end without signing in and without spending a credit.

The per-cell metrics table

Drop a .csv, .tsv, .txt or .md export here, or

choose a file

This is adata.obs as a table — in scanpy, adata.obs.to_csv("qc.csv") after sc.pp.calculate_qc_metrics; in Seurat, write.csv(obj@meta.data, "qc.csv"). Columns are matched by name, not position: total_counts, n_genes_by_counts, pct_counts_mt, pct_counts_ribo, pct_counts_hb, doublet_score, sample, condition, barcode — plus the Seurat aliases (nCount_RNA, nFeature_RNA, percent.mt, orig.ident) and the usual variants. Comma, tab, pipe and semicolon tables all work, as do markdown tables.

Species

Tissue, prep and chemistry

Modality

Cell-type labels

MADs on counts

MADs on genes

MADs on mito %

Hard mito cutoff (%)

Min genes per cell

Doublet score cutoff

Context optional

Paste a metrics table to price the run.

The table, read in your browser

Free, instant, and computed on every pasted row before any run. Every number below is sent to the lane you pick as ground truth, and the worksheet that comes back is reconciled against these flags.

cells—

samples—

median depth—

median mito—

cells kept—

model routed—

Distributions

metric	n	min	q1	median	q3	max	MAD	threshold in force	cells flagged

Per sample

ref	sample	cells	median UMIs	median genes	median mito	mito q3	doublets	removed	retention	conditions	flags

Flags

Conditions across samples

Model routing, before any run

lane

posture

samples

findings

checks passed

steps

Sample by sample

Findings

Named checks

check	status	evidence	what would make it pass

Parameters

name	value	unit	why this value for this dataset

What you do, in order

Every engine flag, reconciled

The browser found these before the run. The worksheet has to deal with each one exactly once — confirming it, or saying why it does not matter for this lane. A flag the worksheet ignored is shown here as ignored.

Assumptions

Open questions

Evidence gaps

In closing

The raw reply

The worksheet did not parse into the structured contract, so here is exactly what came back rather than nothing at all.