Publication [Open access]

If you use SomaticSeq in your work, please cite the following:

  • Fang LT, Afshar PT, Chhibber A, Mohiyuddin M, Fan Y, Mu JC, Gibeling G, Barr S, Bani Asadi N, Gerstein MB, et al. An ensemble approach to accurately detect somatic mutations using SomaticSeq. Genome Biology. 2015;16(1):197. doi:10.1186/s13059-015-0758-2

The SEQC2/MAQC-IV Consortium has published numerous whole-genome and whole-exome sequencing replicates from multiple sequencing centers for a pair of tumor-normal reference samples, along with the high-confidence somatic mutation reference call set. These resources can be used to train machine learning classifers (e.g., Sahraeian SME et al. Genome Biol 2022) or evaluate algorithms and pipelines (e.g., Xiao W et al. Nat Biotechnol 2021). This work is published as:

  • Fang LT, Zhu B, Zhao Y, Chen W, Yang Z, Kerrigan L, Langenbach K, de Mars M, Lu C, Idler K, et al. Establishing community reference samples, data and call sets for benchmarking cancer mutation detection using whole-genome sequencing. Nature Biotechnology. 2021;39(9):1151-1160. doi:10.1038/s41587-021-00993-6 (SharedIt Link).

SomaticSeq was the tool that had Bina Technologies, Inc. ranked #1 and #2 in INDEL and SNV in the Stage 5 of the ICGC-TCGA DREAM Somatic Mutation Calling Challenge.

Documentation can be directly downloaded here.

A quick 8-minute video explaining SomaticSeq

Fang LT, et al. Genome Biol (2015)
SEQC2 somatic mutation reference data and call sets
1st place award 2021 MCBIOS/MAQC Annual Meeting

Fang LT, et al. Nat Biotechnol (2021) / PubMed / SharedIt
How to run SomaticSeq v3.6.3 on precisionFDA

Run in train or prediction mode

Download SomaticSeq

Software Dependencies

Running SomaticSeq

Example command to merge caller results and extract SomaticSeq features

Tumor-normal paired mode

Tumor-only single mode

Additional parameters to be specified before paired or single option to invoke training mode.

Additional input files to be specified before paired or single option invoke prediction mode (to use classifiers to score variants)


Contact Us

Please report bugs and suggestions to the report issues page. The developers will be notified immediately. Alternatively, you may email