Refactor script rules to shell invocations

[iamh2o]

Summary

• replace the script: directives for the count matrix and gtf2bed rules with shell commands that invoke the standalone scripts
• convert the count-matrix, gtf2bed, and gene2symbol helper scripts to accept CLI arguments instead of relying on Snakemake internals
• teach the gene_2_symbol rule to skip annotation on empty inputs and emit a .NODATA marker

Testing

• python -m compileall workflow/scripts/count-matrix.py
• python -m compileall workflow/scripts/gtf2bed.py

---

https://chatgpt.com/codex/tasks/task_e_68cc87ae7af08331aac7d2ade0a75275

[chatgpt-codex-connector]

Codex Review: Here are some suggestions.

Reply with @codex fix comments to fix any unresolved comments.

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, or 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 fix this CI failure" or "@codex address that feedback".

[chatgpt-codex-connector]

[P1] Handle NaN strandedness when building CLI argument string

The new shell invocation builds the --strands argument by joining get_strandedness(units) with commas. get_strandedness returns values straight from the units table, and when the strandedness column exists but contains blanks they come through as NaN floats. Joining a list that contains non‑strings raises TypeError: sequence item 0: expected str instance, float found before the rule even starts, whereas the previous script handled missing values via pd.isnull. This breaks the pipeline for the common case where the column is present but empty in config/units.tsv. Consider normalising to strings (e.g. defaulting NaNs to "none") before joining.

Useful? React with 👍 / 👎.