Bayesian inference for intratumour heterogeneity in mutations and copy number variation

Area of Science:

• Genomics
• Computational Biology
• Cancer Research

Background:

• Tumor samples exhibit heterogeneity due to distinct subclones with varying DNA sequences and copy numbers.
• Accurate variant allele fraction identification relies on precise copy number determination.
• Subclonal copy number and sequence identification is crucial for understanding tumor evolution.

Purpose of the Study:

• To develop a Bayesian feature allocation model for joint subclonal copy number and allele sequence calling.
• To enhance the inference of subclonal structures in tumors.
• To provide a computational tool for analyzing tumor heterogeneity.

Main Methods:

• Utilized a Bayesian feature allocation model with three random matrices for subclonal copy numbers, variant alleles, and cellular fractions.
• Employed next-generation sequencing data for inferring subclonal structures.
• Jointly modeled subclonal structure and sequencing variants for improved posterior inference.

Main Results:

• Demonstrated enhanced posterior inference on subclonal structure through joint modeling.
• Validated the proposed method using simulation studies and real tumor data analysis.
• Showcased the utility of the model in accurately identifying subclonal copy numbers and allele sequences.

Conclusions:

• Joint modeling of subclonal structure and sequencing variants significantly improves the analysis of tumor heterogeneity.
• The developed Bayesian model provides a robust framework for estimating subclonal architecture from sequencing data.
• The associated software (BayClone2) facilitates the application of this method in cancer research.