DeCiFer

DeCiFer infers cancer cell fractions and clusters somatic single-nucleotide variants (SNVs) by modeling mutation multiplicities under the Single Split Copy Number (SSCN) assumption to account for copy-number aberrations (CNAs) and evolutionary mutation losses.

Key Features:

• SSCN-based CCF estimation: Computes cancer cell fractions (CCFs) of SNVs under the Single Split Copy Number (SSCN) assumption, which relaxes the Constant Mutation Multiplicity (CMM) assumption.
• Descendant Cell Fraction (DCF): Implements the Descendant Cell Fraction (DCF) statistic to quantify SNV prevalence while accounting for historical mutation losses due to chromosomal deletions, assigning identical DCFs to SNVs on the same phylogenetic branch.
• Joint multiplicity selection and clustering: Simultaneously selects mutation multiplicities and clusters SNVs based on DCF values to identify SNV clusters corresponding to phylogenetic branches.
• Handling CNAs and mutation losses: Accounts for copy-number aberrations (CNAs) and mutation losses that violate the CMM assumption, improving clustering accuracy in affected tumors.
• Validation and real-data application: Demonstrated improved SNV clustering accuracy in simulations involving mutation losses and applied to sequencing data from 49 metastatic prostate cancer samples producing more parsimonious evolutionary reconstructions.

Scientific Applications:

• Intra-tumor heterogeneity quantification: Enables more accurate quantification of intra-tumor heterogeneity by estimating CCFs and DCFs for SNVs in the presence of CNAs.
• Tumor evolutionary inference: Supports downstream inference and reconstruction of tumor evolutionary histories and phylogenetic branch assignment.
• SNV clustering with CNAs: Clusters somatic SNVs into phylogenetic branches even when SNVs are lost through CNAs.
• Analysis of metastatic prostate cancer: Applicable to sequencing datasets such as the 49 metastatic prostate cancer samples used in published analyses.

Methodology:

Computes CCFs under the SSCN assumption; computes the Descendant Cell Fraction (DCF) for SNVs; simultaneously selects mutation multiplicities and clusters SNVs based on DCF; validated via simulations and applied to sequencing data from 49 metastatic prostate cancer samples.