SynthDNM

SynthDNM classifies de novo single nucleotide polymorphisms (SNPs) and insertions/deletions (indels) by training a random-forest classifier on simulated examples generated from real genome and exome sequencing datasets to enable accurate de novo mutation (DNM) detection across variant calling pipelines.

Key Features:

• Random-forest classifier: Uses a random-forest algorithm to classify potential de novo mutations from variant calls.
• Simulated training data: Constructs simulated training examples from real datasets to generate labeled examples for model training.
• Adaptability across pipelines: Integrates with diverse variant calling pipelines by tailoring simulated training examples to the input data and pipeline-specific outputs.
• Detection of SNPs and indels: Predicts de novo single nucleotide polymorphisms (SNPs) and insertions/deletions (indels).
• Support for genome and exome sequencing: Operates on genome and exome sequencing data and adapts to changes in sequencing technologies and variant calling methodologies.

Scientific Applications:

• Genetic disorder research: Identification of novel de novo mutations that may contribute to inherited or sporadic genetic disorders.
• Population genetics: Analysis of mutation rates and patterns across populations using detected DNMs.
• Precision medicine: Detection of individual-specific de novo variants that can inform clinical genetic interpretation.

Methodology:

Constructs simulated training examples from real genome or exome datasets and trains a random-forest classifier on those examples to classify potential de novo SNPs and indels.