Identifying Lethal Dependencies with HUGE Predictive Power.

Area of Science:

• Genomics
• Cancer Biology
• Computational Biology

Background:

• Functional genomic screens, including CRISPR-Cas9 and RNAi, identify synthetic lethality targets.
• Estimating the effect of genetic events on cell viability reveals lethal dependencies (LEDs).
• The multiple-hypothesis problem in large-scale screens limits statistical power.

Purpose of the Study:

• To improve the prediction of lethal dependencies (LEDs) from functional genomic screens.
• To introduce and validate the “HUb effect in Genetic Essentiality” (HUGE) approach.
• To enhance the discovery of novel, targeted cancer therapies.

Main Methods:

• Analysis of three genome-wide loss-of-function screens (Project Score, CERES, DEMETER).
• Incorporation of the HUGE effect into LED prediction models.
• Validation using cancer disease models (AML, breast, lung, colon) and a clinical interpretation knowledge base.

Main Results:

• HUGE incorporation increased statistical power for LED identification by 75 times compared to state-of-the-art methods.
• Predictions showed high enrichment in clinical interpretations of somatic genomic variants (AUROC > 0.87).
• Identified novel LEDs in acute myeloid leukemia, including FLT3-mutant genotypes sensitive to FLT3 inhibitors.

Conclusions:

• The HUGE approach significantly enhances the power and accuracy of predicting lethal dependencies from functional genomic screens.
• This method facilitates the discovery of novel genetic dependencies for precision-targeted cancer therapies.
• HUGE is effective even in genetically heterogeneous tumors, offering new therapeutic avenues.