Geniml — machine learning on genomic interval data
Train ML models on BED-file genomic intervals with geniml: Region2Vec region embeddings, joint BEDspace embeddings, scEmbed single-cell ATAC-seq embeddings, consensus universe building, tokenization, BEDshift randomization, and BBClient caching
- What
- Train ML models on BED-file genomic intervals with geniml: Region2Vec region embeddings, joint BEDspace embeddings, scEmbed single-cell ATAC-seq embeddings, consensus universe building, tokenization, BEDshift randomization, and BBClient caching
- Cost
- Free
- Needs
- Python with uv (uv pip install 'geniml[ml]' — the [ml] extra pulls torch/gensim) — verified against geniml 0.8.4; external binaries bedtools and uniwig for universe building and tokenization; StarSpace installed separately for BEDspace; scanpy for scATAC-seq workflows; your own BED files
- Install
- Copy the installer prompt below into your Muse — your agent does the rest.
Curated by Skill Harbor — @alterlab-ieu's alterlab-geniml skill, listed here with credit to its creator (part of the AlterLab Academic Skills suite): a complete agent reference for machine learning over genomic interval data with the geniml Python package. It covers Region2Vec (unsupervised word2vec-style embeddings of genomic regions from BED files), BEDspace (joint region + metadata embeddings via StarSpace for cross-modal search), scEmbed (single-cell ATAC-seq cell embeddings that drop into scanpy as `adata.obsm`), consensus peak-set / universe building (`build-universe` with CC/CCF/ML/HMM methods for tokenization references), and utilities (BBClient caching, BEDshift null models, embedding-quality evaluation, tokenization). Notably verified against geniml 0.8.4 — including a hard-won import-path table (subpackage `__init__` files don't re-export, so the obvious imports fail), CLI gotchas (the `bedspace search` query is positional; the StarSpace flag is literally misspelled "starsapce"), and a clear routing table: plain interval arithmetic is gtars, not geniml. Honest caveats: universe building and hard tokenization need the external `bedtools` and `uniwig` binaries; BEDspace needs StarSpace installed separately; universe quality is the whole game — bad universe, bad embeddings; your BED data comes from you. MIT licensed. Skill Harbor never reviews the code, review it yourself before use. Discovered via skills.sh.
Version:
Install
Prerequisites: Python with uv (uv pip install 'geniml[ml]' — the [ml] extra pulls torch/gensim) — verified against geniml 0.8.4; external binaries bedtools and uniwig for universe building and tokenization; StarSpace installed separately for BEDspace; scanpy for scATAC-seq workflows; your own BED files Install "Geniml — machine learning on genomic interval data" for me. It gives my agent @alterlab-ieu's geniml reference: Region2Vec region embeddings, BEDspace joint region+metadata embeddings, scEmbed single-cell ATAC-seq embeddings, consensus universe building (build-universe), tokenization, BEDshift randomization, BBClient caching, and evaluation — with a verified import-path table, CLI gotchas, common workflows (region embedding pipeline, scATAC-seq pipeline, universe building and evaluation), and a routing table for when to use gtars/scanpy/scvi-tools/deepTools instead. Part of the AlterLab Academic Skills suite. MIT licensed. Repository: https://github.com/alterlab-ieu/alterlab-academic-skills/blob/main/skills/domain-specific/alterlab-geniml/SKILL.md 1. Fetch the SKILL.md file (and any helper files) from the repository path into a temporary folder and summarize what it does in one or two sentences. 2. Safety check: review the SKILL.md and scripts for anything suspicious (unexpected network calls, shell commands, credential harvesting). This repo should contain zero secrets in code. Verify that holds here; STOP on any red flag and tell me. 3. Install it as a skill: copy SKILL.md and its helper files into the agent's skills directory, in a folder named "alterlab-geniml". 4. Verify with no network calls: frontmatter valid, files in place. 5. Report what was installed, where, and what I still need to do myself (e.g. uv pip install 'geniml[ml]' plus bedtools/uniwig binaries on PATH; StarSpace separately if using BEDspace; build a good universe first — check tokenization coverage before training). GitHub is optional: if I have a GitHub account or the gh CLI, you may use it; otherwise public access is fine. Never require it unless it's in the prerequisites above. Rules: don't touch anything outside the temp folder and the install target. If anything looks off, stop and ask me.
Questions
How do I install a build?
Every product page includes a copy-paste install prompt. Paste it into your Muse and it sets the build up for you — no manual configuration.
Where does my money go?
Straight to the seller. Skill Harbor never processes payments: checkout happens on the seller’s own page, usually Stripe.
What does the ✓ next to a creator’s name mean?
It means we confirmed the identity of the person behind the listing. It says nothing about the code itself — always check a build before installing it.