#2323rd of 90 in Open modelsFor science, health and research, science
alphagenome
Python client for DeepMind's AlphaGenome DNA models and variant Atlas
- Stars
- 2.2k
- License
- Apache-2.0
- Last commit
- Nov 2026
- Last release
- Nov 2026
- Language
- Python
1,872 on 25 Apr 20262,245 on 14 Nov 2026
From GitHub's daily star counts.
Checked on GitHub.
Overview
The `alphagenome` package is a Python client for the hosted AlphaGenome API; no model weights run locally. It predicts regulatory effects from DNA sequence over 1 Mb intervals and scores variants per cell-type condition. It also queries the AlphaGenome Atlas, which holds pre-computed scores for 9 billion human variants. The API is free for non-commercial use only.
Who it is for: Genomics researchers scoring non-coding variants from Python
Strengths
- Variant scoring with cell-type conditions in a few lines of Python
- Pre-computed Atlas lookups cover 9 billion human variants, also on a no-code website
- Colab tutorials, user guides, vignettes and API reference are published
- Client code is Apache-2.0 and backed by a peer-reviewed Nature paper
Weaknesses
- Needs an API key and network access; inference runs on Google's servers
- Free use is non-commercial only; commercial use goes through Google Cloud
- Live predictions are rate-limited and unsuited to genome-wide analyses
- Outputs may not be used to train other ML models or for clinical decisions
What it needs
- no GPU
- Needs AlphaGenome API key, gRPC, pandas, anndata
- Models: AlphaGenome 2, AlphaGenome 1
What changed
From the release notes. No flag means the note does not say so, not that the upgrade is safe.
Also in Open models
See all 90| Rank | Project | Score |
|---|---|---|
| 1 | ruviewWiFi CSI sensing on ESP32 for presence, breathing, heart rate and pose | 94 out of 100 |
| 2 | layaSingle-pass multilingual classifier returning typed decisions with calibrated odds | 88 out of 100 |
| 3 | rf-detrReal-time transformer models for object detection, segmentation and keypoints | 82 out of 100 |
| #4 | timesfmPretrained time-series foundation model for zero-shot forecasting from Google Research | 80 out of 100 |
| #5 | open_clipPyTorch library for training and running CLIP-style image-text models | 79 out of 100 |