QuadGT

QuadGT performs joint genotype inference across four related human genomes (normal-tumor pairs and their parental genomes) using a general probabilistic Bayesian model to call single-nucleotide variants, de novo germline mutations, and somatic mutations from mapped sequencing reads in standard 1000 Genomes Project file formats.

Key Features:

• Joint Genotype Inference: Integrates variant frequencies from parental genomes with potential de novo germline mutations and somatic mutations to infer genotypes at homologous loci across all four genomes.
• Bayesian Framework: Uses Bayesian inference to quantify descent-by-modification relationships and genotype probabilities.
• Standardized File Formats: Compatible with standard file formats used by the 1000 Genomes Project.
• Algorithms and Data Structures: Implements algorithms and data structures specifically designed for genotype inference in related genome quartets.
• Model Parameter Training: Includes mechanisms for training model parameters based on the input sequencing data.

Scientific Applications:

• Cancer Research: Analyzes normal-tumor genome pairs to identify somatic mutations associated with cancers such as acute lymphoblastic leukemia (ALL), including pediatric cases.
• Genetic Inheritance Studies: Detects de novo germline mutations by leveraging parental genomes to inform inheritance patterns and implications for Mendelian diseases.
• Cohort Exome Analysis: Demonstrated on a cohort of over 120 exomes from childhood ALL quartets to uncover both germline and somatic genetic alterations.

Methodology:

Mapping sequencing reads from quartet samples to the reference human genome, applying a general probabilistic Bayesian model to infer joint genotypes across the four genomes, and using algorithms and data structures with mechanisms for training model parameters.