Viruses and CRISPR · What you can do here
Analyse a pooled CRISPR screen
Analyse pooled CRISPR knockout screens from sgRNA counts using MAGeCK, with gene ranking and honest false discovery rate correction across screening targets.
What this involves
Pooled CRISPR knockout screens allow you to identify phenotypic modifiers and essential genes by measuring sgRNA abundance shifts between experimental conditions. Analysing these experiments requires turning raw read counts into statistically robust gene rankings, accounting for guide variance and multiple hypothesis testing across thousands of targets.
Doing this properly requires strict multiple-comparisons control and complete effect reporting. An uncorrected p-value from a scan of thousands of genes risks false positives; screening hit lists require Benjamini–Hochberg false discovery rate correction to yield valid q-values. In addition, effect estimates, confidence intervals, and replicate numbers must accompany statistical values rather than presenting isolated p-values.
The research agent runs these analyses directly on its dedicated environment where MAGeCK is preinstalled, reading files from your uploads directory and writing outputs to working storage. For broader multi-step workflows, the agent can also execute the nf-core/crisprseq pipeline.
How it goes
- 1Verify environment and inputsThe agent inspects sgRNA count files in /work/uploads/ and verifies the preinstalled MAGeCK installation on PATH without needing setup.
- 2Run MAGeCK analysisMAGeCK processes the sgRNA count tables from /work, contrasting treatment against control replicates using the configured thread count.
- 3Execute background jobs for long runsIf the screen analysis exceeds four minutes, the agent dispatches the process as a background job and monitors completion before retrieving results.
- 4Apply multiple-testing correctionUsing statsmodels in Python, the agent calculates Benjamini–Hochberg q-values across the gene family to separate reliable hits from noise.
- 5Report rankings and statisticsThe agent reports gene effect estimates, confidence intervals, sample counts, and the exact MAGeCK command and parameters used.
Start it here
Ask in plain language. The agent works on a Linux machine of its own — it installs and runs what the job needs, keeps every file, and reads the results back with the numbers explained. Nothing to set up, and a long run continues after you close the page.
What to watch for
- Reporting uncorrected p-values across genome-wide screening targets, which leads to accidental false discoveries.
- Presenting p-values alone without providing effect estimates, 95% confidence intervals, sample size n, and the specific statistical test.
- Interpreting p > 0.05 as 'no difference' rather than reporting that no difference was detected given the observed confidence interval.
- Describing results as 'trending towards significance' or asserting that the screen 'proves' a target without quantitative caveats.
- Failing to state the exact tool command or omitting non-default parameters used during the MAGeCK run.
What it runs
Questions
Do I need to install anything to analyse a pooled crispr screen?
No. The research agent does this work on a Linux machine of its own — it runs MAGeCK there, keeps the files, and reads the results back to you. Long runs continue if you close the page.
Which software performs the CRISPR screen analysis?
MAGeCK 0.5.9.5 is preinstalled on PATH on the machine, and the nf-core/crisprseq pipeline is available for broader automated workflows.
How are multiple comparisons handled for gene hit lists?
The agent applies Benjamini–Hochberg false discovery rate correction using the Python statsmodels library, reporting q-values alongside effect sizes.
Do I need to install MAGeCK or Python packages before running?
No. MAGeCK and Python statistical libraries including scipy, statsmodels, and scikit-learn are already available on the system.
How are long screening runs handled?
Any task taking longer than four minutes is initiated as a background job and checked periodically until output files are written to /work.
Written from the method the research agent follows for this work and the documentation of the tools listed above. Statements citing a figure that does not appear there are removed before publication. Corollary is a non-commercial research demonstration; results are not for clinical use.