Clomial

Clomial infers tumor clonal structure from next-generation sequencing (NGS) data across multiple tumor subsections by fitting a generative binomial model to mutation counts to estimate clonal genotypes and relative frequencies.

Key Features:

• Generative binomial model: Clomial models NGS mutation counts with a generative framework tailored to heterogeneous tumor subsections.
• Expectation-Maximization algorithm: The method uses an expectation-maximization procedure to iteratively estimate clonal genotypes and their relative abundances.
• Binomial fitting to mutation counts: It fits binomial distributions to deep-sequencing mutation counts to quantify variant and normal allele observations.
• Phylogenetic and spatial inference: By comparing mutation frequencies across multiple subsections, Clomial infers phylogenetically coherent and spatially plausible clonal relationships.
• Support for exome and deep sequencing data: The approach operates on exome sequencing followed by deep sequencing of tumor subsections to quantify somatic variant frequencies.

Scientific Applications:

• Oncology research: Inferring clonal architecture to inform studies of cancer evolution, prognosis, and therapy response.
• Primary and metastatic tumor analysis: Application to primary cancers and associated metastases, exemplified by analyses of primary breast cancer and a metastatic lymph node.
• Somatic variant quantification: Quantifying somatic variant frequencies across tumor subsections to elucidate clonal relationships within tumors.

Methodology:

Clomial fits binomial distributions within a generative model and applies an expectation-maximization algorithm to mutation frequency data across multiple tumor subsections to estimate clonal genotypes and relative frequencies.