lamm-mit/skills/depmap

DepMap — Cancer Dependency Map Skill Summary

Overview

This skill enables querying the Cancer Dependency Map project from the Broad Institute to analyze genetic dependencies across cancer cell lines using CRISPR screens, RNAi, and compound sensitivity data.

Primary Use Cases

The skill supports identifying cancer-selective gene dependencies, validating oncology drug targets, discovering synthetic lethal interactions, and uncovering biomarkers that predict treatment sensitivity.

Core Data Types

Dependency Scores:

• Chronos (CRISPR): ranges ~-3 to 0+, where more negative indicates higher essentiality
• Standard thresholds: ≤-0.5 suggests dependence; ≤-1 indicates strong dependence
• Gene Effect: normalized version where -1 represents median effect of common essential genes

Cell Line Information: Each line includes unique DepMap ID, name, primary disease classification, tissue lineage, and lineage subtype.

Technical Implementation

The skill provides Python-based access through:

1. RESTful API endpoints at https://depmap.org/portal/api/
2. Direct data downloads from https://depmap.org/portal/download/all/
3. Local analysis of CSV files including gene effect matrices, mutation data, copy number, and expression

Key Analytical Workflows

Target Validation: Filter cell lines by cancer type and compute selective dependency patterns for candidate genes.

Synthetic Lethality: Compare gene effect scores between mutant and wild-type cell lines to identify selective dependencies.

Biomarker Discovery: Correlate genomic features (mutations, expression) with dependency scores using statistical testing.

Co-Essentiality: Identify genes with correlated dependency profiles suggesting shared pathways or complexes.

Critical Best Practices

• Prioritize current Chronos scores over legacy DEMETER2 data
• Distinguish broadly essential genes (poor drug targets) from cancer-selective dependencies
• Validate findings against expression data since unexpressed genes appear non-essential
• Account for copy number artifacts in essential gene calls
• Apply multiple-testing correction for genome-wide analyses