Tutorials

Worked analyses, start to finish.

A public dataset, the prompts exactly as they were typed, the decisions as they came up, and the plots that carried the conclusion. Every tutorial links the full generated report, so you can check the write-up against the run.

Bulk RNA-seqMechanism of action

What is dexamethasone doing to airway smooth muscle?

Paired differential expression on an FPKM matrix that was never meant for it, a cross-check against the submitters' own pipeline, a claim withdrawn under verification, and a blind test that asks the signature to name its own drug.

GSE52778 · 4 donors, dex vs untreated, 8 samples~20 min read

Read the tutorial

Principal component analysis separating dexamethasone-treated from untreated airway smooth muscle samples

Single-cell RNA-seqMechanism of action

What is IFN-β doing to PBMCs, cell type by cell type?

24,562 cells collapsed into 108 donor-paired libraries, a lineage ranking on a scale cell number cannot inflate, an interaction test that finds the shared core thirty times smaller than the specific remainder, and a QC filter the run says did nothing.

GSE96583 · ~25 min read

Mass-spec proteomicsBiomarker discovery

Which proteins mark response to melanoma immunotherapy?

Zero proteins at FDR in any cohort, a metadata audit that removes a third of the deposition, a directional story that turned out to be a sign error the run found in itself, and a purity adjustment that keeps one programme and cannot separate the other from immune infiltrate.

PXD006003 · ~30 min read

How these are built.

Each tutorial is one analysis session, not a curated highlight reel. The prompts are quoted as they were typed, including the ones that were vague. Where Inflexa proposed a design choice, the screenshot of that proposal is on the page, because the choice is the interesting part and a walkthrough that hides it is a demo.

Where a run left something unresolved, the tutorial says so and says why it was left. An analysis that reads as frictionless is less useful than one that shows judgment being applied, and it is also less true.

Run one of these yourself.

The datasets are public and the prompts are on the page. Install Inflexa and point it at the same accession.

Install Inflexa