CNVPipe: An enhanced pipeline for accurate analysis of copy number variation from whole-genome sequencing

Copy number variations (CNVs) are critical contributors to the genetic architecture of complex diseases, yet many existing pipelines for whole-genome sequencing (WGS) data exhibit persistently high false discovery rates (FDR). Here, we introduce CNVPipe, an enhanced workflow that integrates widely used CNV-calling tools with a novel machine-learning framework to achieve lower FDR and higher sensitivity. CNVPipe also provides a specialised module for single-cell CNV analysis, featuring a new model that accurately estimates the ploidy of single cells to get integer copy number profile. Benchmark evaluations demonstrate that CNVPipe outperforms current pipelines in various genomic contexts. In addition to detecting recurrent CNVs in paediatric developmental disorders, CNVPipe enables large-scale functional genomics applications involving stem cell technologies. Moreover, its application to sparse single-cell DNA sequencing data provides a new aspect of research in cancer evolution. Collectively, these findings underscore the versatility and reliability of CNVPipe as a comprehensive solution for CNV detection in WGS-based research.