dpCNV

dpCNV detects copy number variations (CNVs) from next-generation sequencing (NGS) read depth (RD) profiles using a density peak-based statistical approach to identify significant genomic bins as CNVs.

Key Features:

• Density Peak Clustering: Utilizes local density and minimum distance metrics from RD-derived features to identify significant clusters corresponding to CNV events.
• Two-Dimensional Data Representation: Transforms local density and minimum distance features into a two-dimensional representation for downstream statistical analysis.
• Significance Testing: Constructs a two-dimensional null distribution for genome bins to test bin-level significance and flag potential CNVs.
• Read Depth Feature Extraction: Extracts local density and minimum distance features directly from sequencing read depth profiles.
• Data Types Supported: Applies to both simulated datasets and real sequencing samples.
• Performance Metrics: Demonstrates improved sensitivity and F1-score in empirical evaluations relative to comparator methods.

Scientific Applications:

• Complex Disease Research: Detection of CNVs implicated in complex diseases using NGS read depth data.
• Genome Mutation Analysis: Supplemental CNV identification in studies of genomic variation and mutation across simulated and real datasets.

Methodology:

Implements a density peak clustering algorithm using local density and minimum distance features extracted from RD profiles, maps these features to a two-dimensional representation, and builds a two-dimensional null distribution to test the significance of each genome bin.