Association analysis using somatic mutations.

Area of Science:

• Genomics
• Cancer Research
• Statistical Genetics

Background:

• Somatic mutations are key drivers of tumor growth and cancer treatment biomarkers.
• Standard regression methods are inadequate for somatic mutation association studies due to false positive/negative mutation call rates.
• Advancements in sequencing necessitate new statistical methods for large-scale somatic mutation analysis.

Purpose of the Study:

• To develop and implement a novel statistical method, SAME (Somatic mutation Association test with Measurement Errors), to address mutation calling uncertainty.
• To assess the associations between somatic mutations and various outcomes, including gene expression and cancer subtypes.
• To improve the statistical power and accuracy of somatic mutation association studies.

Main Methods:

• Developed the Somatic mutation Association test with Measurement Errors (SAME), a likelihood-based approach.
• Implemented SAME with computationally efficient software.
• Validated SAME through extensive simulation studies and application to The Cancer Genome Atlas (TCGA) data.

Main Results:

• SAME significantly improves statistical power compared to generalized linear models that ignore mutation calling uncertainty.
• SAME identified novel associations between somatic mutations and gene expression across 12 cancer types.
• SAME revealed associations between somatic mutations and colon cancer subtypes, uncovering findings missed by conventional methods.

Conclusions:

• SAME provides a robust and statistically powerful approach for somatic mutation association studies.
• The method effectively handles uncertainty in somatic mutation calls, leading to more reliable results.
• Mutation-level and gene-level analyses are differentially appropriate for oncogenes and tumor-suppressor genes, respectively.