subHMM

subHMM infers allele-specific copy number alterations (ASCNA), subclone genotypes, and clonal proportions from tumor genomic data using a hidden Markov model to characterize tumor subclonal architecture.

Key Features:

• ASCNA and subclone region identification: Identifies genomic regions harboring distinct tumor subclones and allele-specific copy number alterations.
• Genotype and clonal proportion estimation: Estimates subclone genotypes and their clonal proportions across the genome.
• Two-step parameter estimation: Fits a standard hidden Markov model using a conglomerated hidden state for clonal genotype and subclone status, then derives region-specific clonal proportions by maximizing pseudo-likelihoods.
• Statistical inference methods: Employs the Expectation-Maximization algorithm and the Forward-backward algorithm for parameter estimation and state decoding within the HMM framework.
• Validation: Validated by simulation studies demonstrating accurate identification and characterization of subclone regions and genotypes in analyses of somatic copy number alterations.

Scientific Applications:

• Tumor heterogeneity and evolution studies: Characterizes subclonal architecture to support analyses of tumor progression and heterogeneity.
• TCGA renal cell carcinoma analyses: Applied to renal cell carcinoma datasets from The Cancer Genome Atlas (TCGA) to study subclones and somatic copy number alterations in real-world genomic data.

Methodology:

Fits a hidden Markov model with a hidden state variable representing combined clonal genotype and subclone status, uses the Expectation-Maximization algorithm and the Forward-backward algorithm for parameter estimation and state inference, and refines region-specific clonal proportions by maximizing pseudo-likelihoods.