SMURFS: A comprehensive Nextflow pipeline for consensus somatic variant detection

Accurate somatic variant identification remains a fundamental challenge in cancer genomics, where algorithmic variability and technical artifacts can significantly compromise variant calling accuracy. Single-caller approaches often lead to high false positive rates and poor reproducibility, while implementing multi-caller consensus strategies requires complex bioinformatics integration and standardized quality control frameworks. Here, we present SMURFS (Somatic MUtation Recognition Framework & Suite), a comprehensive Nextflow pipeline that addresses these challenges through ensemble variant calling, rigorous quality control, and additional integration of copy number and structural variant detection. SMURFS implements four complementary variant calling algorithms-Mutect2, SAGE, Strelka2, and MuSE2-with consensus calling by at least two independent callers, reducing potential false positive mutations while maintaining high sensitivity for true somatic variants. The pipeline incorporates critical quality control measures including contamination detection and tumor-normal concordance verification by Conpair, Mosdepth coverage profiling, and DKFZBiasFilter systematic artifact identification. Beyond point mutations, SMURFS integrates ASCAT for allele-specific copy number profiling, CNVkit for total copy number assessment, and Manta/Delly for structural variant detection, providing comprehensive somatic alteration characterization. The pipeline supports flexible analysis entry points accommodating raw FASTQ files, aligned BAMs, duplicate-marked BAMs, or recalibrated BAMs, with compatibility across four reference genomes (GRCh38, GRCh37, mm39, RN7) covering over 80% of cancer genomics research scenarios. Performance test using TCGA whole-exome sequencing samples demonstrated an average runtime of 277.8 minutes at a cost of 17.33 per sample pair on AWS m4.16xlarge instances. SMURFS addresses the reproducibility crisis in somatic variant calling by providing a standardized, accessible framework that combines state-of-the-art algorithms with comprehensive quality control. The modular architecture enables customization for specific research requirements while maintaining analytical rigor. This unified framework represents a significant advancement in cancer genomics methodology, enabling more reliable identification of driver mutations, mutational signatures, and therapeutic targets across diverse cancer types.