Ten strongest selective dependencies
Ranked by most-negative MSI − MSS mean difference among genes with Benjamini–Hochberg adjusted P < 0.05. Click a gene to explore its models.
| Rank / Gene | MSI n | MSS n | MSI mean | MSS mean | MSI − MSS | Adjusted P |
|---|
More-negative Chronos scores indicate stronger growth impairment. The release scales common nonessential genes around 0 and common essential genes around −1; these are reference points, not hard dependency thresholds. Tests are two-sided Welch tests with correction across all successfully tested genes.
From group difference to individual cell lines
Every point is one model. Select a point or a source-table row to inspect its original value.
The summaries above are recalculated from the displayed source rows. Filters change the plot, table and summaries; the ranked results and their P-values always refer to the full cohort. Solid vertical ticks show displayed group means; dashed vertical line marks 0.
Source values
| ModelID | Cell-line name | Lineage | Group | MSI score | Exact gene effect |
|---|
Original decimal strings are retained in the table and source CSV. Gene-results CSV contains the full-cohort result and the current filtered summary; it does not attach full-cohort P-values to a filtered comparison.
Provenance, exclusions and checks
Sources and original-file checksums
DepMap 24Q4 Public, version 1 · Broad DepMap (2024) · CC BY 4.0. No newer live-release values are used.
| Source | Bytes | Original MD5 |
|---|
Complete originals were downloaded and verified against the versioned Figshare metadata. SHA-256 checksums are also included in the downloadable provenance record.
Group assignment, joins and exclusions
Metadata-only and signature-only models have no usable gene-effect row and cannot enter this analysis. Missing model labels do not remove otherwise usable observations; absent labels would display as Unknown.
Methods and interpretation limits
Each model contributes once by ModelID. No lineage adjustment, covariate adjustment, within-patient clustering or causal attribution is applied. Filtering is descriptive; it does not rerun the genome-wide tests or multiple-testing correction.
| Lineage | MSI n | MSS n |
|---|
Independent value verification and sensitivity checks
| Gene | MSI median | MSS median | 10% trimmed Δ | Δ after removing 3 lowest MSI |
|---|
These descriptive sensitivity checks help distinguish broad shifts from extreme observations. They do not alter the ranking or replace the prespecified Welch tests. The three-model check removes only the three most-negative MSI values and keeps MSS unchanged.