Correlate

Loading gene effect data...

Reference Data

Loading data...

Data from DepMap 25Q3:

CRISPRGeneEffect.csv (gene effects)
Model.csv (cell line metadata)
OmicsSomaticMutationsMatrixHotspot.csv (hotspot mutations)
OmicsFusionFiltered.csv (translocations/fusions)

Please acknowledge DepMap if you use this data.

Open

1. Set Parameters

Analysis Mode:

Analysis (within input genes) Design (all genes) Mutation Analysis (differential GE) Synonym/Ortholog Lookup

Correlation Cutoff: 0.50

Min Slope: 0.10

Analysis Sub-type:

Hotspot Mutation Translocation/Fusion

Hotspot Mutation (required):

Min Cell Lines:

Lineage Filter:

Subtype Filter:

Exclude Tissues: ▼ click to expand

Hotspot Mutation Filter:

Selecting for mutations may introduce bias — mutations are often enriched in specific cancer types.

Translocation/Fusion Filter:

2. Input Genes

Gene symbols (one per line):

Enter Gene, LFC, FDR (tab or comma separated): Stats auto-loaded when you click Run

3. Results

New to Correlate?

Click Test Genes and then Run ▶

(both in "2. Input Genes")

Next, double-click nodes and edges in the generated network for more functions. Decrease correlation cutoff and press Run again to identify interactions of weaker strength.

Tip: Click Correlate to view a scatter plot, or By Tissue for tissue breakdown. Ctrl/Cmd+click a column header to copy that column. Use Filters to set numeric cutoffs on columns.

↗

Gene1	Gene2	Correlation	Slope	N	Cluster	Analyze
Run analysis to view correlations

Mean (All) = all cells, Mean (Filt) = filtered cells. * indicates genes from your input list (Design mode). All input genes are shown; "Corr" column indicates if correlations were found.

↗

Gene	Cluster	Mean Effect	SD
Run analysis to view clusters

Note: Differential gene effects may reflect selection bias (e.g., mutations enriched in certain cancer types) rather than direct functional consequences of the mutation. Consider filtering by lineage to control for tissue-specific effects.

Statistics: p-values are calculated using Welch's t-test comparing gene effect scores between wild-type (WT, 0 mutations) and mutated cells (1+2 or 2 mutations). Δ GE = Mean(mutated) − Mean(WT).

0 genes with p < 0.001

↗

	Gene	N (WT)	Mean GE (WT)	N (1+2)	Mean GE (1+2)	Δ GE ▲	p-value	N (2)	Mean GE (2)	Δ GE (2v0)	p-value (2v0)	N (Fused)	Mean GE (F)	Δ GE (F)	p (F)

Low Risk: Exact synonyms from HGNC. Mid Risk: Less certain synonyms. Orthologs: Mouse-to-human gene mappings.

Input Gene	Status	Official Symbol	Low Risk Synonyms	Mid Risk Synonyms	Orthologs (Mouse→Human)
Select "Synonym/Ortholog Lookup" mode and click Run

Run analysis to view summary

Gene Effect:

Min n: View:

Statistics (click row to focus)

p-values: Welch's t-test comparing each cancer type vs all other cell lines.

Group	N	Mean GE	SD

Export Chart: Export CSV:

What is Correlate?

Correlate analyses gene effect scores (CRISPR knockout fitness) across 1,100+ cancer cell lines from the DepMap project. Genes whose knockout produces correlated fitness effects across cell lines are likely to share functional relationships — shared pathways, protein complexes, or regulatory dependencies.

Understanding Gene–Gene Correlation

Each gene pair is compared by their gene effect scores across all cell lines. The Pearson correlation (r) measures how similarly two gene knockouts affect cell fitness:

↗

Positive (r > 0)

Knockout of both genes affects the same cell lines in the same way — suggesting shared pathway or complex membership.

⟷

No correlation (r ≈ 0)

Gene knockouts have unrelated fitness effects across cell lines.

↘

Negative (r < 0)

Gene knockouts have opposite effects across cell lines — when one is essential, the other is dispensable, suggesting compensatory or antagonistic roles.

TSC1 vs TSC2 scatter plot showing strong positive correlation (r=0.90)

Example: TSC1 vs TSC2 gene effect across 1,186 cell lines (r = 0.90). Knockout of either gene produces nearly identical fitness effects, yet the impact is highly context-dependent — some cell lines thrive while others are depleted.

TSC1/TSC2 signalling pathway

The strong correlation reveals that TSC1 and TSC2 function as an obligate complex — without prior knowledge of the pathway. This is the power of unbiased correlation analysis: it discovers functional relationships directly from the data.

Pathway adapted from Huang & Manning, Cell Res (2008) 18:62–69

The Correlation Network

All pairwise correlations between your genes are combined into an interactive network graph. Each node is a gene; edges connect correlated pairs. Edge color and thickness encode the correlation:

Correlation network showing gene relationships

Example network: Genes cluster by functional relationships. TSC1–TSC2 (thick blue edge) share a complex; TP53–MDM2 (red edge) show opposing fitness effects.
Double-click a node to see its gene effect profile, or an edge to view the scatter plot.

From Screen Data to Hypotheses

Correlate is designed for two complementary purposes: analysing screen data (e.g. CRISPR hits, gene lists) and generating new hypotheses using its built-in tools:

🔍 Design Mode

Enter seed genes and discover what correlates with them. Expand beyond your input list to find unexpected functional partners.

🧬 Mutation Analysis

Compare gene effect in cells with specific hotspot mutations or translocations vs wild-type. Identify genes with differential dependency in mutant backgrounds.

📊 Cell Line Browser

Explore individual cell lines — filter by tissue, mutation, or gene effect. View top dependencies and select cell line subsets for focused analyses.

🧪 Enrichr Integration

Send any gene list to Enrichr for pathway enrichment analysis. Available from correlations, clusters, mutation results, and cell line views.

Quick start: Click Test Genes, then Run ▶ to see an example analysis.
Lower the correlation cutoff and re-run to reveal weaker interactions.