Genetic mapping and predictive modeling of paralog synthetic lethality

Area of Science:

• Genomics
• Systems Biology
• Cancer Biology

Background:

• Paralogous genes are abundant in the human genome and are a potential source of synthetic lethality.
• The human paralogome remains largely uncharacterized, limiting its therapeutic potential.

Purpose of the Study:

• To investigate the frequency and penetrance of synthetic lethality in paralogous gene pairs.
• To identify factors contributing to the variable penetrance of paralog synthetic lethality across different cancer types.

Main Methods:

• A large-scale digenic screen of 36,648 paralogous pairs in the human genome.
• Machine learning classification applied to paralog pairs across 49 cancer models.
• Predictive modeling of synthetic lethal interactions based on protein-protein interaction networks.

Main Results:

• Synthetic lethalities were infrequent and showed variable penetrance across different tumor backgrounds.
• Endogenous perturbations in related pathways predicted paralog synthetic lethality.
• The strength of synthetic lethal interactions correlated with the overlap and essentiality of shared protein-protein interaction networks.

Conclusions:

• The heterogeneity of paralog synthetic lethality is influenced by complex polygenic interactions and cellular contexts.
• Understanding protein-protein interaction network properties is crucial for predicting synthetic lethal interactions.
• This study provides a framework for characterizing the paralogome and exploiting synthetic lethality in cancer therapy.