The MERGE (Model Ensemble for Ranking Genomic Effect) framework is a pipeline for variant pathogenicity prediction. By avoiding reliance on clinical databases that cause "data circularity," MERGE uses a Dynamic Shunting Architecture that routes every variant to a dedicated panel rather than scoring everything with one general-purpose model.
There are four panels — coding, splice, non-coding and non-SNV (indel / MNV / delins) — each with its own feature set and its own classifier, trained separately. The panel is chosen automatically: a variant whose REF or ALT is not a single base goes to the non-SNV panel; otherwise ANNOVAR annotation decides between coding, splice and non-coding.
| Panel | Classifier | Features |
|---|---|---|
| Coding | Logistic regression (L1) | 6 |
| Splice | CatBoost | 10 |
| Non-coding | Logistic regression (L2) | 10 |
| Non-SNV | CatBoost | 10 |
The panels draw on up to 14 model features from genomic foundation models and protein language models: AlphaGenome, Enformer, Evo2 (7B, 7B-base and 1B-base), Carbon-3B, GENERATOR and GENERATOR-v2, Nucleotide Transformer and NT-v2, HyenaDNA, AlphaMissense, ESM-1b and GPN-MSA. Any feature that cannot be computed for a given variant is median-imputed by the ensemble, and every prediction reports which features were computed and which were imputed.
The MERGE software is open source under the MIT licence. The web server is
provided free of charge for research use. By using this website or the
merge-cli command-line tool, you agree to the following.
merge-cli source code carries no
field-of-use restriction — commercial use, modification and redistribution are all
permitted. See
github.com/TheFangLab/MERGE.Required AlphaGenome notice.
The AlphaGenome API is provided for non-commercial use only and is subject to the AlphaGenome Terms of Service. Outputs generated by AlphaGenome should not be used for the training of other machine learning models.
Predictions are for theoretical modelling and research purposes only; they should not be used for clinical decision-making or relied upon for medical or other professional advice.
For collaboration or any question about these terms, contact the team using the addresses below.
merge-cli is the professional-user counterpart to this website: the same models and the same ensemble, scriptable and runnable on your own machine. It is published on PyPI and, is published under the MIT licence — the source carries no field-of-use restriction. The four bundled ensemble model files are CC BY-NC 4.0; see the terms above.
It covers all four variant panels — coding, splice, non-coding and non-SNV (indel / MNV / delins) — and up to 14 model features (AlphaGenome, Enformer, Evo2-7B / 7B-base / 1B-base, Carbon-3B, GENERATOR, GENERATOR-v2, Nucleotide Transformer, NT-v2, HyenaDNA, AlphaMissense, ESM-1b, GPN-MSA).
Requires Python 3.11 or newer.
Scores are computed by this server; the MERGE ensemble runs locally on your CPU. The variant panel (coding / splice / non-coding / non-SNV) is chosen automatically.
Local mode runs every model on your own hardware — no rate limits and no variant ever leaves your machine. Each model is deployed into its own conda environment and served over HTTP; deploy only the ones you need.
Scoring windows and conventions are fixed to the values the ensemble was trained on, so results match the published model. A model you do not run is simply median-imputed by the ensemble, and the CLI always reports which features were computed and which were imputed.
merge-cli ships an MCP server, so assistants such as Claude Desktop, Claude Code and Codex can call MERGE directly. Both steps are needed: the first registers the tools, the second makes the assistant reach for them.
The MCP server always runs in remote mode, so no GPU is required. Full documentation ships with the package
(TUTORIAL.md) and on the
PyPI project page.