Add comprehensive CLI reference and skill documentation - #4
Conversation
Adds root-level SKILL.md and REFERENCE.md so Claude can use maxATAC as a Skill: constructing correct maxatac CLI commands, understanding required inputs/outputs for each subcommand, and chaining prepare -> predict/train/benchmark workflows, without duplicating the existing docs/readme/*.md content.
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 4da73ccac6
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| maxatac predict -tf CTCF --signal GM12878_IS_slop20_RP20M_minmax01.bw -o outputdir/ | ||
| maxatac predict -tf CTCF --signal GM12878_IS_slop20_RP20M_minmax01.bw --roi ROI.bed | ||
| maxatac predict -tf CTCF --signal GM12878_IS_slop20_RP20M_minmax01.bw --chromosomes chr3 chr5 |
There was a problem hiding this comment.
When users copy these predict examples, they will fail before running because maxatac/utilities/parser.py registers -n/--name/--prefix for predict with required=True. The reference then lists the prefix as required below, so the examples should include something like -n GM12878_CTCF; otherwise the new reference's primary predict workflow is not runnable.
Useful? React with 👍 / 👎.
| ChIP-seq-derived gold-standard bigwig (1 = TFBS, 0 = no TFBS). | ||
|
|
||
| ```bash | ||
| maxatac benchmark --prediction GM12878_CTCF_chr1.bw --gold_standard GM12878_CTCF_ENCODE_IDR.bw --chromosomes chr1 --bin_size 200 |
There was a problem hiding this comment.
In the benchmark parser the prediction input is only accepted via the mutually exclusive -bed/--bed or --bw/--bigwig/-bw options, and a prefix is required via -n/--name/--prefix; there is no --prediction flag. Copying this example will therefore hit argparse's required input/prefix errors instead of benchmarking, so update it to use --bw ... -n ... (or the BED equivalent).
Useful? React with 👍 / 👎.
| |---|---| | ||
| | `--genome` | Genome build. | | ||
| | `-m`, `--model` | Trained `.h5` model. | | ||
| | `-i`, `-s`, `--signal` | ATAC-seq signal bigwig. | |
There was a problem hiding this comment.
For maxatac variants, -s is already the sequence option, while the signal bigwig is defined as -i, -bw, -signal, or --signal in the parser. Documenting -s as the signal flag will make commands generated from this table either fail the required input_bigwig check or pass the bigwig path as a sequence file, so remove -s from this row and keep it only under sequence.
Useful? React with 👍 / 👎.
Restructure into the skills/<name>/SKILL.md convention instead of a root-level SKILL.md.
2 participants
Summary
This PR adds two new documentation files to provide comprehensive guidance on using the maxATAC CLI toolkit: a detailed command reference and a skill definition for AI assistants.
Changes Made
REFERENCE.md: Complete CLI reference documentation covering all 11
maxatacsubcommands (data,prepare,average,normalize,predict,train,threshold,peaks,benchmark,variants). For each subcommand, includes:SKILL.md: AI assistant skill definition providing:
Notable Details
-n/--name/--prefix) to help users discover correct syntax-slop 20default that must match for pretrained model compatibilityhttps://claude.ai/code/session_015q4D4NyYsATVTN3V3kiMrt