Skip to content

Add query-genomic-intelligence skill - #46

Open
boldakov wants to merge 2 commits into
Runchuan-BU:mainfrom
genomicintelligence:add-query-genomic-intelligence
Open

Add query-genomic-intelligence skill#46
boldakov wants to merge 2 commits into
Runchuan-BU:mainfrom
genomicintelligence:add-query-genomic-intelligence

Conversation

@boldakov

Copy link
Copy Markdown

Add query-genomic-intelligence skill

Adds a runtime skill wrapping Genomic Intelligence (GI), a set of hosted
transformer DNA language models that predict regulatory features, gene structure,
and expression directly from DNA sequence. No local GPU or model weights are
needed; the skill is a thin client over a hosted, versioned inference API.

Per CONTRIBUTING.md, new capabilities arrive as skills. This is a runtime skill
under container/skills/, modelled on query-ensembl: a single SKILL.md with
inline requests.

What it does

Give it a gene symbol, a genomic region, or a DNA/FASTA sequence and it returns
structured predictions across six tasks, plus a composite:

  • promoter: promoter regions
  • splice: donor/acceptor splice sites
  • enhancer: developmental and housekeeping activity
  • chromatin: chromatin state across hundreds of tracks
  • expression: sequence to expression, log(TPM+1)
  • annotation: de-novo gene/transcript annotation
  • composite: find the genes in a region and predict each one's expression

How it connects

The REST /v1 API needs a GI_API_KEY bearer and is called as
POST /v1/tasks/{task}/predict. The hosted MCP server at
https://mcp.genomicintelligence.ai/mcp works keyless against a public demo
quota, with the key optional.

The skill documents both paths, the Ensembl-based sequence acquisition (including
the exact 9,198 bp TSS-centred window the expression model requires), and the
async annotation poll.

Scope

This is a skill-only change. It adds one file,
container/skills/query-genomic-intelligence/SKILL.md, and touches no source
files, so the skills-only PR check passes. Verified with
git diff --name-only upstream/main...HEAD, which lists that path alone.

No secrets are committed; the key is read from the GI_API_KEY environment
variable. The keyless MCP path was verified against the live endpoint: an
anonymous Streamable HTTP session with no Authorization header completes
initialize and returns a real predict_promoter result.

Notes

For research and development use, not clinical or diagnostic decisions.

Docs: https://docs.genomicintelligence.ai

boldakov added 2 commits July 23, 2026 23:23
Wrap Genomic Intelligence's hosted DNA-sequence models (promoter, splice,
enhancer, chromatin, expression, annotation, plus a find-genes-then-expression
composite) over the REST /v1 API and the keyless hosted MCP server. Single
SKILL.md with inline requests, mirroring the query-ensembl skill. Skill-only
change, no source files touched.
fetch_ensembl_sequence takes gene (fetch_region handles coordinates),
load_demo_sequence requires name, and find_genes /
find_genes_and_predict_expression take a sequence handle rather than a region.
The annotation task is find_genes; there is no predict_annotation and no
load_local_fasta on the hosted server. Also drop dangling references/errors.md
pointers, add the 422 validation_failed case, and state the 9,198 bp window
consistently.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant