Evo 2 Skill

Evo 2 is a long-context DNA language model, and this is the operating guidance for it: likelihood scoring, embeddings, generation, and variant comparison. It answers questions about the sequence itself — how likely is this base, this window, this edit. When the question is instead what an assay would have measured, that is borzoi, and a two-axis variant prioritization runs both. The model code, the checkpoints, and the accelerator runtime are all external.

GPU capacity, usable context length, checkpoint access, and generation quality depend on the environment you actually run in, so confirm them there instead of assuming them from this page. And a likelihood is a model score: it is not a measured variant effect.

Install

A Skill is a directory of files, so installing one is copying that directory somewhere an agent looks. With Node 18+ and git on PATH (npm resolves a github: spec through it), and nothing cloned:

npx github:PKU-YuanGroup/OpenAI4S install evo2 --target claude

--target claude writes to ~/.claude/skills, claude-project to ./.claude/skills, openai4s to <data_dir>/user-skills, and --dir <path> to anywhere you name. The resolved absolute path is printed before anything is written there, and --dry-run stops at that plan. A reinstall refuses to overwrite a copy you have edited or one it did not install, and uninstall removes only the files it wrote. For a recipe included in the npm release, npx @pku-yuangroup/openai4s-skills install evo2 --target claude installs the published copy. The npm catalog can differ from this repository.

Without Node, take the directory itself — and turn it into a .zip if an upload field wants one. The tarball is the whole repository (over 100 MB), and the pipe is a POSIX shell recipe (macOS, Linux, WSL) that extracts only this directory:

curl -L https://codeload.github.com/PKU-YuanGroup/OpenAI4S/tar.gz/refs/heads/main \
  | tar -xz --strip-components=2 OpenAI4S-main/skills/evo2
python3 -m zipfile -c evo2.zip evo2

The click-through form of that same download is the repository zip, which is also the route to take on Windows: extract it and copy skills/evo2/ out (unzip main.zip 'OpenAI4S-main/skills/evo2/*' unpacks only that directory). If you already run OpenAI4S there is nothing to install — the wheel ships every bundled Skill, and a bundled Skill takes precedence over a same-named copy in <data_dir>/user-skills. Targets, provenance, and what the installer refuses to do: tools/skills-installer/.

Files

File Responsibility

SKILL.md Sequences go in as plain list[str]; hand score_sequences a tensor and it dies on a dtype mismatch, because the API tokenizes for you. It gives back one mean log-likelihood per sequence, and a variant is scored as Δll = ll_alt - ll_ref over a fixed window. generate returns .sequences, .logits and .logprobs_mean, all populated with no flag to set. Embedding genomic windows is in the declared scope too. Then the model table — 7B in roughly 22 GB, 40B in roughly 78 GB, both at a million nucleotides of context — the remote-compute path for those checkpoints, and the failure modes worth recognizing early, including an HF_HOME on a read-only mount that only breaks when the loader tries to write refs/.

File	Responsibility
`SKILL.md`	Sequences go in as plain `list[str]`; hand `score_sequences` a tensor and it dies on a dtype mismatch, because the API tokenizes for you. It gives back one mean log-likelihood per sequence, and a variant is scored as `Δll = ll_alt - ll_ref` over a fixed window. `generate` returns `.sequences`, `.logits` and `.logprobs_mean`, all populated with no flag to set. Embedding genomic windows is in the declared scope too. Then the model table — 7B in roughly 22 GB, 40B in roughly 78 GB, both at a million nucleotides of context — the remote-compute path for those checkpoints, and the failure modes worth recognizing early, including an `HF_HOME` on a read-only mount that only breaks when the loader tries to write `refs/`.