Comprehensive Evaluation of Major Analysis Parameters for Somatic Whole Exome Sequencing

Our study aims to evaluate and optimize Next-Generation Sequencing (NGS) variant calling pipelines by investigating how different parameters affect pipeline performance, accuracy, and computational efficiency.

We tested 96 NGS pipelines using 2 mapping algorithms (BWA, Bowtie) and 3 variant callers (Strelka, Somatic Sniper, Mutect) with different parameters (‘isTrimmed’, ‘baseRecalibration’, ‘markDuplicates’) across 2 computing environments and 5 centers, producing 480 variant lists.

We benchmarked these pipelines to assess their performance on accuracy, execution time, and stability. Our analysis found that the choice of parameters like trimming and base recalibration significantly affects execution time and result stability. The combination of BWA and Mutect with ‘isTrimmed’ and ‘baseRecalibration’ set to yes provided the most stable and accurate results.

The project includes a detailed comparison of parameter combinations, highlighting trade-offs between accuracy and computational efficiency.