evo2
GitHub基于Evo2基因组大模型，提供DNA序列打分、嵌入及生成能力，用于变异效应评估和调控区域分析。
Trigger Scenarios
Install
npx skills add aipoch/open-science --skill evo2 -g -y
SKILL.md
Frontmatter
{
"name": "evo2",
"license": "Apache-2.0",
"category": "biomodels",
"metadata": {
"third_party": [
{
"kind": "weights",
"name": "Evo 2",
"license": "Apache-2.0",
"provider": "Arc Institute",
"terms_url": "https:\/\/github.com\/ArcInstitute\/evo2\/blob\/main\/LICENSE"
}
],
"display-name": "Evo 2"
},
"description": "Score, embed, and generate DNA sequences with Evo 2, a long-context genomic foundation model. Use this skill when: (1) Computing per-nucleotide or per-sequence likelihoods for variant effect\n scoring,\n(2) Embedding genomic windows for downstream classification, (3) Generating DNA conditioned on a prefix, (4) Scoring regulatory or coding regions across species.\n",
"requirements": [
"gpu"
]
}
Evo 2 — DNA Language Model
Prerequisites
| Requirement | Minimum | Recommended |
|---|---|---|
| Python | 3.11 | 3.12 (<3.13) |
| CUDA | 12.1+ | 12.4+ |
| GPU VRAM | 24 GB (7B bf16) | 80 GB (40B) |
| RAM | 32 GB | 128 GB |
How to run
Installation
pip install evo2
# Weights pulled from Hugging Face on first model load.
Loading and scoring
from evo2 import Evo2
model = Evo2("evo2_7b") # or "evo2_40b" — see model table
seqs = ["ATCG" * 50, "GGGCTTAA" * 25]
ll = model.score_sequences(seqs) # → list[float], mean per-token log-likelihood
print(ll)
Generation
out = model.generate(
prompt_seqs=["ATGAAAGCT"],
n_tokens=256,
temperature=0.7,
)
print(out.sequences[0])
Models
| Name | Params | Context | VRAM (bf16) | Notes |
|---|---|---|---|---|
evo2_7b |
7 B | 1 M nt | ~22 GB | Default; fits on a single 24 GB+ GPU |
evo2_40b |
40 B | 1 M nt | ~78 GB | H100 80 GB or multi-GPU |
evo2_1b_base |
1 B | 8 K nt | ~6 GB | FP8 path requires sm_89+ (H100) |
Output format
score_sequences returns a list[float] (or np.ndarray) of mean log-likelihoods,
one per input sequence. More negative ⇒ less likely under the model. For variant
effect, compute Δll = ll_alt - ll_ref over a fixed window.
generate returns a GenerationOutput with .sequences (list[str]), .logits
(list[Tensor]), and .logprobs_mean (list[float]) — always populated, no flag required.
Decision tree
Need a DNA model?
│
├─ Per-base/per-sequence likelihood, generation → Evo 2 ✓
├─ Predict experimental tracks (expression, accessibility) → borzoi
└─ Protein, not DNA → fair-esm2 / esmfold2
Remote compute
7B/40B inference is GPU-bound (≥24 GB / 80 GB VRAM). Read
compute_details({provider, mode:'read'}) for an environment with evo2 +
flash-attn and a pre-cached HF weight mount, then submit:
c = host.compute.create(provider)
job = c.submitJob(
intent="Evo2-7B score 200bp variant window — 1×GPU, ~2 min",
inputs=[{"src": "score_evo2.py", "dstFilename": "score_evo2.py"}],
command="python3 score_evo2.py", # env selection is host-specific — see compute_details for your provider
outputs=["scores.json"],
timeoutSeconds=1800,
)
print(job.job_id) # cell ends here — kernel never blocks on compute
Retain the exact returned job_id. Query that saved ID with the non-blocking
c.attachJob(job_id).status() or .result() when its state or result is relevant; do not scan Job
history. A final .result() read reports whether its follow-up was suppressed or had already been
committed; otherwise the app starts the later analysis turn for an unread final result. See the
remote-compute-ssh skill for details.
Inside score_evo2.py, point HF_HOME at the provider's weight-cache mount
(path is in compute_details) and set HF_HUB_OFFLINE=1 so the loader
doesn't try to write refs/ into a read-only mount. Weight footprint:
~15 GB (7B), ~80 GB (40B).
Typical performance
| Task | 7B on H100 | Notes |
|---|---|---|
| Model load (cached) | ~5-7 min | First call hydrates weights |
score_sequences, 200×200bp |
~10-20 s | After load |
generate, 1×512 nt |
~15 s |
Troubleshooting
| Symptom | Cause | Fix |
|---|---|---|
Transformer Engine not installed |
No FP8 — falls back to bf16 | Informational only on non-H100; ignore |
| OOM on load | 40B on <80 GB GPU | Use evo2_7b or shard with device_map |
HF tries to write refs/main |
HF_HOME points at RO mount |
Set HF_HUB_OFFLINE=1 |
dtype mismatch in score_sequences |
Passing tensors not strings | Pass list[str]; the API tokenises for you |
Next: pair with borzoi to predict track-level effects of the same
variants.
Version History
- 44394f0 Current 2026-09-11 11:10
