ifCNV

ifCNV detects copy number variations (CNVs) in next-generation sequencing (NGS) datasets using isolation-forest machine learning to identify CNV-positive samples and generate an internal reference for analysis.

Key Features:

• Isolation Forest-Based Algorithm: Employs an isolation-forest-based machine learning algorithm to detect CNV signals as outliers in read-depth data.
• Self-Referencing Capability: Generates an internal reference framework from the analysed sample set, avoiding reliance on an external normal dataset.
• Dual Isolation Forests with Comprehensive Scoring: Integrates two isolation forests combined with a comprehensive scoring method to improve sensitivity and specificity of CNV calls.
• Validation Across Diverse Data Types: Validated on multiple dataset types including capture and amplicon data and on both germline and somatic samples.
• High Sensitivity, Specificity, and Accuracy: Demonstrates high sensitivity, specificity, and accuracy for CNV detection across varied NGS contexts.

Scientific Applications:

• Genetic Research: Facilitates detection of CNVs to support studies of genetic variation and genotype–phenotype associations.
• Personalized Medicine: Supports identification of CNVs relevant to patient stratification and individualized therapeutic decisions.
• Oncology: Enables detection of somatic CNVs that inform tumor profiling and potential therapeutic targets.
• Clinical Genomic Analysis: Applicable to clinical workflows requiring robust CNV identification from NGS data.

Methodology:

Uses isolation-forest-based machine learning with two isolation forests and a comprehensive scoring method, plus automatic identification of CNV-positive samples and internal reference generation from the sample set.