Release notes from deepsomatic

DeepSomatic 1.10.0

2026-03-05T23:37:06Z

DeepSomatic:

Steamlining input parameters: run_deepsomatic now reads parameters from model.example_info.json files which must be present with the models to run.
Small model in DeepSomatic: Introduced small models for tumor-normal modes in DeepSomatic improving the runtime between 12% to 40%.

Contributions:

Ehud Amitai (@ehudamitai) from Ultima genomics for the algorithm development of multiallelic variant post-processing method that is available as “product” option.
Vasiliy Strelnikov (@vaxyzek) for streamlining the run_deepvariant script by enabling automatic flag loading using model.example_info.json files.
Sowmiya Nagarajan (@sonagarajan) - for helping to update the RNA-seq model.
Shezan Rohinton Mirzan (@shezanmirzan) for migrating small model to Keras 3 and modernizing core infrastructure.
Francisco Unda (@fcoUnda) for enhancing read sampling stability, fixing non-determinism, and creating robust read sampling approach at high coverages.
Alec Zhang (@az-e) for providing essential internal updates and maintenance to the codebase.

DeepSomatic 1.9.0

2025-05-13T19:12:24Z

DeepSomatic:

In this release, we are introducing FFPE_WGS_TUMOR_ONLY and FFPE_WES_TUMOR_ONLY models.
The WGS and WGS_TUMOR_ONLY models have been retrained with all datasets described in the manuscript, tumor-in-normal and normal contamination datasets.
Overall, we see improved generalization because of training dataset updates. We highly recommend updating to 1.9.0 for DeepSomatic analysis.

Detailed release note for DeepVariant can be found in: https://github.com/google/deepvariant/releases/tag/v1.9.0

DeepSomatic 1.8.0

2025-04-02T18:26:50Z

In this release:

Updated the doc to detail the code availability.
No model updates.

[Updated on 2025-04-02] We are updating the WGS tumor-normal models for both r1.7.0 and r1.8.0 to ensure robustness across some rare somatic events across different cancer types.

r1.8.0

2024-12-03T23:23:52Z

DeepSomatic 1.8 release

DeepSomatic 1.7.0

2025-04-02T18:26:54Z

Updates included in this release:

DeepSomatic now has tumor-normal models to support WES and ONT.
DeepSomatic now has FFPE-trained models for FFPE_WGS and FFPE_WES.
DeepSomatic now supports tumor-only variant calling with Illumina WGS, ONT and PacBio.
Improved accuracy for tumor-normal variant calling with Illumina WGS and PacBio.
Models are trained with varying purities that improve somatic calling accuracy in cases of tumor-normal cross contamination.
Improved runtime in make_examples and postprocess variants through code optimization.
Models are trained with extensive training data described in this manuscript.

We are sincerely grateful to:

Juan Carlos Mier (@jmier2) for his code and PacBio model training contributions.
Jimin Park (@jimin001) (UCSC), Benedict Paten (@benedictpaten) (UCSC) and Mikhail Kolmogorov (@fenderglass) (NIH, NCI) for testing and providing analysis results.
Jorge Gonzalez (@jgonzalezmendez) for helping to improve runtime.
PacBio: Khi Pin Chua (@proteinosome), William Rowell (@williamrowell) for feedback on PacBio somatic calling.

[Updated on 2025-04-02] We are updating the WGS tumor-normal models for both r1.7.0 and r1.8.0 to ensure robustness across some rare somatic events across different cancer types.

DeepSomatic v1.6.1

2024-03-19T19:23:08Z

In this release:

We fixed a bug in call_variants that caused the step to freeze in cases where there were no examples. This bug was observed and reported in google/deepvariant#764, google/deepvariant#769, #8.
Updated libssw library from 1.2.4 to 1.2.5.
The same model files are used for v1.6.0 and v1.6.1 for all technologies.

DeepSomatic 1.6.0

2023-10-24T05:12:23Z

Initial Release of DeepSomatic (v1.6)
DeepSomatic versioning will attempt to keep cadence with DeepVariant.
Pre-trained models for Illumina and Pacific Biosciences data available.
We are grateful to Khi Pin Chua, William Rowell (@williamrowell) and Aaron Wenger for helping us develop the PacBio model.
We are also grateful to Jimin Park (@jimin001) (UCSC), Benedict Paten (@benedictpaten) (UCSC) and Mikhail Kolmogorov (@fenderglass) (NIH, NCI) for testing and providing analysis results.