Free case sampleReal public dataNo account

Would you sign off this RNA-seq result?

An analyst has sent a result table and a draft caption. Decide what you would accept, revise or ask about before opening the explanation.

Review the handoff Download guided sample

This sample is free. The complete case is under review and is not available for purchase.

Your first decision

Would you use this caption?

“TSPAN6 increases after dexamethasone because its log2 fold change is positive.”

The attached table is labelled untreated relative to treated. It contains unshrunken estimates from a paired model; adjusted p values retain the full analysis’s correction.

Received excerpt · untreated relative to treated
Gene	log2 fold change	Adjusted p
TSPAN6	0.385	9.60e-4
DPM1	-0.203	0.184
FKBP5	-4.043	7.44e-26

Write a decision and one sentence of evidence in your own notes. You do not need R or a matrix-rank calculation. Then open the walkthrough to compare your reasoning.

The results come from public data; the handoff is an authored teaching scenario. This is a self-check, not a secure assessment. The walkthrough and guided download contain explanations. No response is submitted or saved by this page.

Open the guided walkthrough and feedback

01 / Inspect the evidence

Same fit. Opposite comparison.

Choose a gene, then reverse the comparison. The browser re-expresses a precomputed result; it does not rerun DESeq2.

Four cell lines, each measured twice

log2(size-factor-normalized count + 1)

N61311N052611N080611N061011

The axis rescales between genes. Points stay fixed when you reverse the comparison. These normalized counts retain cell-line differences; they are not batch-corrected values.

TSPAN6 / Treated vs untreated

-0.385

log2 fold change · unshrunken estimate

For TSPAN6, expression is estimated lower in treated relative to untreated cells.

Model fold ratio: 0.766×
Standard error: 0.101
Two-sided p value: 1.28e-4
BH adjusted p value: 9.60e-4

Meets this analysis’s adjusted-p-value threshold of 0.05. This does not establish clinical benefit.

Inspect the eight values and sample IDs

TSPAN6 · size-factor-normalized counts, rounded for display
Sample	Cell line	Condition	Count
SRR1039508	N61311	Untreated	664.4
SRR1039509	N61311	Treated	500.0
SRR1039512	N052611	Untreated	743.6
SRR1039513	N052611	Treated	610.4
SRR1039516	N080611	Untreated	968.2
SRR1039517	N080611	Treated	748.3
SRR1039520	N061011	Untreated	840.1
SRR1039521	N061011	Treated	605.2

02 / Make the review decision

Three short decisions. Check an answer to see its reasoning, then try again if you wish. This is practice, not a certification test.

1 / 3A handoff gives TSPAN6 a positive log2 fold change for untreated relative to treated. Does that support ‘TSPAN6 increases after treatment’?

Yes. A positive fold change always means treatment increases expression.No. In that comparison, positive means higher in untreated cells.The adjusted p value alone determines whether expression went up.

2 / 3Reverse the comparison in this fixed fit. What should happen to the two-sided p value?

It changes sign with the log2 fold change.It becomes one minus the old p value.It stays the same; the effect sign reverses.

3 / 3DPM1 has a positive treated-relative-to-untreated estimate, but its adjusted p value is about 0.184. Which conclusion is supported at 0.05?

The estimate is positive, but DPM1 does not meet this threshold.DPM1 definitely does not respond to treatment.A positive estimate proves that treatment helps patients.

Where these numbers come from

This educational reanalysis uses the eight libraries in airway 1.32.0: four human airway smooth-muscle cell lines measured with and without dexamethasone. It is a subset of GSE52778, from Himes and colleagues (2014).

The reference run used DESeq2 1.52.0, raw counts, ~ cell + dex, and a two-sided Wald test. It retained 16,139 of 63,677 genes using a rule of at least ten counts in at least four libraries. BH adjustment used the full eligible family after independent filtering at 0.05. These three display genes were chosen for teaching; the fit and adjustment were not run on a three-gene subset.

Effects are unshrunken estimates. Normalized counts are for inspection, not a replacement for model input. The example does not recheck read-level quality, establish a biological mechanism, or demonstrate clinical benefit. See the DESeq2 method documentation for assumptions and extensions.

Download the displayed data (JSON) · Upstream attribution · LGPL notice. Data reuse follows the upstream package’s declared LGPL terms; no exclusive rights to the public data are claimed.

Beyond the sample

Review the whole handoff.

The complete case candidate adds sample correspondence, repeated measurements and a separate study plan to review. The required output is your own memo, with evidence for each decision. Running R and inspecting matrix rank are optional extensions.

Count matrix and a deliberately reordered sample sheet
A runnable reference analysis with recorded outputs
A memo template, worked explanation and self-review rubric
A new scenario to test whether the same reasoning transfers

In review · no enrolment or payment is open

No release date, live sessions, personal data analysis or ongoing support is promised. Existing free resources remain available.

Request a study feasibility audit