Predict tissue and cell-type expression from TSS-centered DNA sequences via the Genomic Intelligence API.
npx skills add https://github.com/ClawBio/ClawBio --skill gi-expression
Please help me install this Agent Skill. Skill: gi-expression Source: https://github.com/ClawBio/ClawBio/tree/main/skills/gi-expression Command: npx skills add https://github.com/ClawBio/ClawBio --skill gi-expression
Sequence-to-expression models like Enformer, Borzoi, and G0 Expression normally require GPU infrastructure, private model weights, and careful 9-kbp windowing around transcription start sites. This Skill removes that burden by calling the hosted Genomic Intelligence expression prediction endpoint, returning predicted log TPM and TPM values conditioned on a free-text cell-type or assay description.
Ask the agent to run the gi-expression demo to predict HBB expression in K562 cells, or provide your own TSS-centered FASTA file and cell-type description for a custom prediction.
High-intent search queries and answers about installing and using this skill.
Provide a single-record FASTA file containing a 9,198 bp window centered on the transcription start site, plus a cell-type description. The skill posts it to the Genomic Intelligence /v1/tasks/expression/predict endpoint and returns predicted log TPM and TPM in a Markdown report and JSON result.
The API accepts sequences from 9,198 to 500,000 bp. Submit exactly one 9,198 bp TSS-centered window, or a longer locus with the --tss-index flag giving the 0-based TSS offset so the API cuts the scored window server-side.
Yes, it requires a Genomic Intelligence partner key set via the GI_API_KEY environment variable or the --api-key flag. A shared hackathon-tier key ships in the repo's .env.example, and individual keys can be requested from Genomic Intelligence for heavier use.
The most common cause is wrong strand orientation: minus-strand genes must be reverse-complemented to gene-sense before submission. An incorrect --tss-index also returns a confident but wrong result, so always check the scored window line in report.md.
TPM predictions are useful as relative rankings within a cell type, not as absolute cross-tissue measurements. Also keep the description phrasing fixed across comparisons, since paraphrasing the free-text conditioning string can change predictions by roughly a factor of two.
Do not use it when you already have RNA-seq count data and want differential expression analysis; that requires a dedicated RNA-seq DE workflow. It is also unsuitable for clinical or diagnostic decisions, as predictions are model outputs rather than measurements.
pradyumnasagarCompute reverse complements and complements for DNA/RNA sequences with IUPAC ambiguity support.
justaddcoffeeGuide genomics data analysis decisions for RNA-seq, microarray, and single-cell data.
K-Dense-AIPredict promoters, splice sites, enhancers, chromatin state, and gene expression from DNA sequences via hosted transformer models.
HughYauPredict genome-wide functional tracks from DNA sequence using the Borzoi deep learning model.
