CaMutQC: An R package for integrative quality control and filtration of cancer somatic mutations

Abstract

The quality control and filtration of cancer somatic mutations (CAMs), including the elimination of false positives due to technical bias and the selection of key mutation candidates, are crucial steps for downstream analysis in cancer genomics. However, due to diverse needs and the lack of standardized filtering criteria, the filtering strategies applied vary from study to study, often resulting in reduced efficiency, accuracy, and reproducibility. Here, we present CaMutQC, a heuristic quality control and soft-filtering R/Bioconductor package designed specifically for CAMs. CaMutQC enables users to remove false positive mutations, select potential mutation candidates, and estimate Tumor Mutation Burden (TMB) with a single line of code, using either default or customized parameters. A filter report and a code log can also be generated after the filtration process to facilitate reproducibility and comparison. The application of CaMutQC to a Whole-exome Sequencing (WES) benchmark dataset demonstrated its strong capability by eliminating 85.55 % of false positive Single nucleotide variants (SNVs) while retaining 90.72 % of true positive SNVs. Additionally, an additional 11.56 % of true positive SNVs were rescued through CaMutQC's built-in union strategy. Similar results were observed for Insertions and Deletions (INDELs). CaMutQC is freely available through Bioconductor at https://bioconductor.org/packages/CaMutQC/ under the GPL v3 license.

Keywords: CaMutQC; Cancer somatic mutation; False positive mutation; Mutation filtering; Quality control.

Graphical abstract

Fig. 1

A. An overview of CaMutQC. B. CaMutQC connects the upstream and downstream of variant calling by taking annotated variant calls, filtering them, and passing them to downstream tools.

Fig. 2

Benchmark results of WES_EA_1. Broken lines were used at the top of the y-axis for better displaying the main data range. A. TP and FP SNV counts after different procedures. The number of FP SNVs has greatly decreased after CaMutQC filtration with/without internal filtration. B. TP and FP INDEL counts after different procedures. raw: raw counts with no filtration; C: filtered by CaMutQC (CaMutQC-mutFilterTech); internal: processed by callers’ internal filter.

Fig. 3

Benchmark results of sample WGS_EA_1. A. TP and FP SNV counts after different procedures. B. TP and FP INDEL counts after different procedures. C: filtered by CaMutQC (CaMutQC-mutFilterTech); internal: processed by callers’ internal filter.

Fig. 4

An example of the updated union strategy. A. An illustration of the union strategy applied in this study. The union of CaMutQC-filtered (CaMutQC-mutFilterTech) variants from three callers, MuTect2, VarScan2, and MuSE, was obtained. B. Union results of SNVs in WES_EA_1. The number of TPs increases by 18.36 % on average among the three callers.