Inferring clonal composition from multiple sections of a breast cancer

Area of Science:

• Genomics
• Computational Biology
• Oncology

Background:

• Cancers develop through mutation and selection, creating diverse clonal populations.
• Assessing tumor clonal composition is crucial for prognosis and treatment.
• Inferring clonal structure from next-generation sequencing (NGS) data is challenging.

Purpose of the Study:

• To develop a generative model and expectation-maximization algorithm for deconvolving NGS data to infer tumor clonal structure.
• To estimate clonal genotypes and their frequencies from multi-subsection tumor data.

Main Methods:

• Proposed a generative model for NGS data from multiple tumor subsections.
• Developed an expectation-maximization algorithm for estimating clonal genotypes and frequencies.
• Validated the approach using simulations and applied it to a primary breast cancer and lymph node metastasis.

Main Results:

• The algorithm successfully inferred phylogenetically and spatially plausible clonal relationships.
• Quantified frequencies of 17 somatic variants in a breast cancer sample.
• Demonstrated the validity of the generative model and algorithm via simulation.

Conclusions:

• The developed method accurately infers tumor clonal composition from NGS data.
• This approach can elucidate cancer clonal evolution across space and time.
• Future application to more tumors will enhance understanding of cancer development.