Quantifying gene selection in cancer through protein functional alteration bias

Area of Science:

• Genomics
• Computational Biology
• Cancer Research

Background:

• Identifying cancer driver genes is crucial for understanding and treating cancer.
• Existing computational methods for driver gene discovery rely on assumptions about mutation distributions.
• There is a need for robust methods to identify cancer-associated genes with minimal assumptions.

Purpose of the Study:

• To present FABRIC, a novel framework for quantifying gene selection in cancer.
• To assess the effects of de novo somatic mutations on protein-coding genes.
• To identify cancer genes with statistically significant biases towards harmful mutations.

Main Methods:

• Utilized a machine-learning model to quantify functional effects of approximately 3 million somatic mutations from over 10,000 human cancer samples.
• Compared observed mutation effects against all possible single-nucleotide mutations in the coding human genome.
• Developed FABRIC to compare each gene against its own background model using rigorous statistics, minimizing assumptions about random somatic mutation distributions.

Main Results:

• Detected 593 protein-coding genes exhibiting a statistically significant bias towards harmful mutations.
• Discovered genes show substantial overlap with known cancer genes, while also highlighting previously overlooked candidates.
• FABRIC framework demonstrated effectiveness in avoiding false discoveries through rigorous statistical comparison.

Conclusions:

• FABRIC provides a robust, assumption-light framework for identifying cancer driver genes.
• The identified genes offer new insights into cancer mechanisms and potential therapeutic targets.
• FABRIC is an open-source tool with a user-friendly command-line interface, facilitating broader adoption in cancer research.