elastic-net-feature-selection

1# Elastic Net Feature Selection
2 
3## When to Use
4 
5- Use this skill for binary case-vs-control classification on bulk expression matrices.
6- Use it when you need elastic net logistic regression feature selection, coefficient paths, and `cv.glmnet`-based lambda selection.
7- Use custom labels such as `Tumor` and `Normal` only when the group file still contains exactly two outcome levels.
8 
9## Out of Scope
10 
11- Survival or Cox modeling
12- Multiclass outcomes
13- Single-cell data
14- Non-expression tables
15 
16Out-of-scope enforcement:
17 
18- If the group file contains any label outside the requested `case_group` and `control_group`, the command stops with `SKILL_INVALID_DATA` instead of silently dropping samples.
19- If either requested class is missing after validation, the command stops with `SKILL_INVALID_DATA`.
20 
21## When to Read External Files
22 
23| Situation | File to Read | Purpose |
24|-----------|--------------|---------|
25| **Need to understand alpha, lambda choice, or feature-selection behavior** | `references/algorithm.md` | Elastic net logistic regression, penalty mixing, cross-validation, and coefficient selection assumptions |
26| **Need the authoritative executable entrypoint** | `scripts/main.R` | Run: `Rscript scripts/main.R --input_file ... --group_file ... --output_dir ...` |
27| **Need parameter examples, smoke-test commands, or recorded local runs** | `references/cli-guide.md` | Verified CLI examples for normal runs, conservative runs, and test-data runs |
28…