PEcnv

PEcnv detects copy number variations (CNVs) in panel sequencing data using an Exponentially Weighted Moving Average (EWMA)-based statistical process model to improve sensitivity for small CNVs (1 kb-1 Mb).

Key Features:

• Statistical Process Model: Employs a statistical process model to mitigate issues including optimal sliding window setting and correction for bias and noise in CNV detection.
• Exponentially Weighted Moving Average (EWMA): Calculates regional read depths using an EWMA strategy to provide adaptive smoothing of sequencing read-depth variability.
• Dynamic Sliding Window: Uses a self-adaptive sliding window whose size adjusts based on weighted averages derived from the EWMA to detect CNVs from kilobase-scale to chromosome-arm level.
• Bias/Noise Reduction Model: Integrates a model with the moving average approach to reduce bias and noise, manage complex patterns, and extend training data for improved accuracy, particularly for small CNVs (1 kb-1 Mb).
• Performance Validation: Validated on simulated and real samples with comparative analyses demonstrating improved detection of small CNVs in panel sequencing data relative to existing methods.

Scientific Applications:

• Clinical testing with panel sequencing: Enables detection of clinically relevant small CNVs that are often missed by methods optimized for larger CNVs.
• Research on genetic variation and complex disease: Facilitates identification of CNVs from targeted sequencing panels for studies of genetic contributors to complex diseases.

Methodology:

Computes regional read depths with EWMA, applies a self-adaptive sliding window sized from EWMA-derived weighted averages, and integrates a bias/noise reduction statistical process model.