PyLOH

PyLOH resolves the identifiability problem in heterogeneous tumor NGS data by integrating somatic copy number alterations (SCNAs) and loss of heterozygosity (LOH) within a probabilistic framework to identify tumor-associated reads and estimate tumor purity and ploidy.

Key Features:

• SCNA and LOH integration: Integrates somatic copy number alterations (SCNAs) and loss of heterozygosity (LOH) in a unified probabilistic framework.
• Identifiability resolution: Addresses the identifiability problem where different combinations of tumor purity and ploidy can explain NGS data equally well.
• Read-level assignment: Identifies reads associated with specific tumor cells or subclones within mixed samples.
• Purity and ploidy estimation: Estimates tumor purity and ploidy from mixed tumor-normal next-generation sequencing samples.
• Heterogeneous sample analysis: Handles heterogeneous tumor samples containing mixtures of normal and multiple clonal tumor cell populations.
• Model-derived algorithms: Implements algorithms derived from the probabilistic model to improve inference accuracy.
• Benchmarking: Benchmarked on simulated data and 12 breast cancer sequencing datasets, reporting significant improvements over existing methods in resolving identifiability and estimating tumor purity.
• Implementation: Implemented in Python.

Scientific Applications:

• Purity–ploidy disambiguation: Resolving tumor purity–ploidy identifiability in cancer genome sequencing analyses.
• Subclone read assignment: Assigning sequencing reads to tumor subclones for analyses of intra-tumor heterogeneity.
• Tumor–normal deconvolution: Deconvolution and analysis of mixed tumor and normal cell populations in NGS datasets.
• Method validation: Validation and benchmarking of SCNA- and LOH-based inference using simulated and breast cancer sequencing datasets.

Methodology:

Integrates SCNAs and LOH within a unified probabilistic model and applies algorithms derived from that model to identify tumor-associated reads and estimate tumor purity and ploidy; validated on simulated data and 12 breast cancer sequencing datasets; implemented in Python.