CNV-PG

CNV-PG performs copy-number variant (CNV) prediction and genotyping from paired-end sequencing data using machine-learning classifiers to reduce false positives and integrate calls from multiple CNV callers.

Key Features:

• CNV-P (prediction): Predicts CNVs from paired-end sequencing data using classifiers trained on validated subsets from various CNV callers.
• CNV-G (genotyping): Produces CNV genotypes and is compatible with existing CNV calling tools.
• Machine-learning framework: Employs machine-learning classifiers trained on validated subsets from multiple CNV callers to distinguish true CNVs from false positives.
• False-positive reduction: Filters false-positive CNVs generated by existing discovery algorithms to improve precision.
• Integration and unification: Integrates CNV calls from multiple callers into a unified, high-confidence call set.
• Implementation: Implemented in Python for computational CNV prediction and genotyping.

Scientific Applications:

• Genetic disorder research: Provides more accurate CNV calls for studies investigating the role of CNVs in genetic disorders.
• Complex trait and disease genomics: Facilitates analysis of CNV contributions to the genetic architecture of complex traits and conditions.

Methodology:

Uses machine-learning classifiers trained on validated subsets from various CNV callers to refine CNV predictions and genotypes from paired-end sequencing data, integrates calls from multiple callers, and filters false positives; implemented in Python.