Sparse dictionary learning recovers pleiotropy from human cell fitness screens

Area of Science:

• Genomics
• Systems Biology
• Computational Biology

Background:

• High-throughput functional genomic screens typically assume gene products have singular functions.
• However, gene perturbations can cause multiple outcomes, a phenomenon known as pleiotropy.
• Understanding pleiotropy is crucial for interpreting complex genetic data.

Purpose of the Study:

• To develop a computational model for dissecting pleiotropy in large-scale fitness screen data.
• To infer independent functional effects from single gene perturbations.
• To uncover complex genetic architectures underlying high-dimensional biological readouts.

Main Methods:

• Developed a novel computational approach (Webster) to model pleiotropy.
• Represented gene perturbations as sums of multiple functional perturbations with independent fitness effects.
• Inferred fitness effects empirically from experimental data.

Main Results:

• Successfully recovered pleiotropic functions for DNA damage proteins in genotoxic screens.
• Untangled distinct signaling pathways upstream of shared proteins in cancer cell screens.
• Predicted protein complex stoichiometry using only fitness data.
• Identified compound mechanisms of action by modeling compound sensitivity profiles.

Conclusions:

• The Webster approach effectively models pleiotropy in high-throughput genomic screens.
• This method provides a powerful tool for unraveling complex genetic architectures.
• It enables deeper insights into gene function, pathway interactions, and drug mechanisms.