QuickSeg: A fast, versatile and accurate algorithm for genomic copy number segmentation using dynamic programming

Copy number alterations are prevalent in cancer genomes, and their accurate identification hinges on precise segmentation of read-depth signals from sequencing data. Current segmentation methods prioritize computational efficiency over accuracy, often falling short due to sensitivity to technical artifacts in sequencing data. In response, researchers introduce QuickSeg, an innovative and efficient approach that employs an exact dynamic programming algorithm to detect copy number segments utilizing a median-based error function.

Observing the presence of a small but significant population of outlying observations in sequencing depth distributions, QuickSeg offers enhanced robustness against technical noise while simultaneously streamlining the computational complexity of segmentation tasks.

In extensive evaluations of whole-genome sequencing data from cancer cohorts, employing breakpoint-supported somatic copy number alterations, QuickSeg demonstrates superior segmentation precision compared to two established baseline methods, Circular Binary Segmentation (CBS) and Piecewise Constant Fitting (PCF), across a wide range of sensitivity thresholds.

Furthermore, QuickSeg consistently outperforms both methods in terms of runtime and memory usage. The findings collectively demonstrate the biological and computational advantages of robust median-based optimization for copy number segmentation, enabling accurate analysis of extensive sequencing cohorts with minimal computational requirements.