Characterizing Intra-Tumor Heterogeneity From Somatic Mutations Without Copy-Neutral Assumption

Area of Science:

• Genomics
• Computational Biology
• Cancer Research

Background:

• Tumor samples are inherently heterogeneous, containing diverse cancer cell subpopulations (subclones).
• Each subclone possesses a distinct mutational genotype profile, crucial for understanding cancer progression.
• Accurately resolving tumor heterogeneity is vital for effective cancer treatment strategies.

Purpose of the Study:

• To introduce ChaClone2, a novel computational method for deconvolving tumor heterogeneity.
• To accurately estimate subclonal genotypes, cellular proportions, and the number of subclones within a tumor.
• To account for the influence of copy number aberrations on mutation loci.

Main Methods:

• Utilizes a state-space formulation of a feature allocation model.
• Employs an efficient sequential Monte Carlo (SMC) algorithm for parameter estimation.
• Deconvolves variant allele fractions (VAFs) from patient samples into subclonal copy numbers and proportions.

Main Results:

• ChaClone2 demonstrates superior accuracy compared to existing state-of-the-art methods.
• The method exhibits scalability for analyzing large cancer genomics datasets.
• Model parameter estimates can be updated with new mutation data.

Conclusions:

• ChaClone2 provides an accurate and scalable solution for deconvolving tumor heterogeneity.
• The method facilitates a deeper understanding of cancer progression and informs treatment decisions.
• Availability of MATLAB code and datasets supports further research and application.