PhenoDriver: interpretable framework for studying personalized phenotype-associated driver genes in breast cancer

Area of Science:

• Oncology
• Genomics
• Bioinformatics

Background:

• Identifying personalized cancer driver genes is crucial for understanding cell transformation and clinical diagnosis.
• Existing methods often focus on gene identification at cohort or individual levels but lack mechanism elucidation.
• A gap exists in connecting driver genes to their specific oncogenic roles and clinical phenotypic alterations.

Purpose of the Study:

• To introduce PhenoDriver, an interpretable framework for personalized cancer driver gene identification.
• To elucidate the oncogenic mechanisms of identified driver genes and their association with clinical phenotypes.
• To discover novel breast cancer subtypes and their underlying molecular mechanisms.

Main Methods:

• Developed and applied the PhenoDriver framework to analyze genomic data from 988 breast cancer patients.
• Evaluated PhenoDriver's performance against state-of-the-art methods for cohort-level driver gene identification.
• Utilized PhenoDriver to identify both recurrent and rare driver mutations in individual patients and construct gene subnetworks.

Main Results:

• PhenoDriver demonstrated superior performance in identifying breast cancer driver genes at the cohort level.
• The framework effectively identified personalized driver genes, including those with rare mutations.
• Investigated oncogenic mechanisms of known and novel driver genes (e.g., TP53, MAP3K1, HTT) and their associated subnetworks.
• Discovered two known and one novel breast cancer subtype with elucidated molecular mechanisms.

Conclusions:

• PhenoDriver provides a robust method for personalized cancer driver gene identification and mechanism discovery.
• The findings enhance the understanding of breast cancer mechanisms and offer insights for therapeutic decisions.
• The framework aids in developing targeted anticancer therapies by revealing personalized driver profiles and subtypes.