Update version to 0.18.1 and enhance public API formatting

[parashardhapola]

• Bump package version to 0.18.1.
• Refactor all in init.py for improved readability.
• Update gene ID regex in validation.py to allow for versioned gene IDs.
• Modify _id_like_percentage function to support random sampling with a seed for better statistical representation.

[qodo-code-review]

Review Summary by Qodo

Version 0.18.1: API formatting and gene ID validation improvements

✨ Enhancement

Walkthroughs

Description

• Bump package version to 0.18.1
• Refactor __all__ list formatting for improved readability
• Update gene ID regex to support versioned identifiers
• Enhance _id_like_percentage with random sampling and seed parameter

Diagram

flowchart LR
  A["Version Update<br/>0.18.0 → 0.18.1"] --> B["Public API<br/>Formatting"]
  A --> C["Gene ID<br/>Validation"]
  C --> D["Support Versioned<br/>Gene IDs"]
  C --> E["Random Sampling<br/>with Seed"]
  B --> F["Enhanced<br/>Readability"]

Loading

File Changes

1. cytetype/__init__.py ✨ Enhancement +7/-2

Version bump and public API list formatting

• Bumped version from 0.18.0 to 0.18.1
• Reformatted __all__ list with multi-line formatting for improved readability

cytetype/init.py

---

2. cytetype/preprocessing/validation.py ✨ Enhancement +10/-4

Gene ID validation and statistical sampling enhancements

• Added random module import for sampling functionality
• Updated gene ID regex pattern to support optional version suffixes (e.g., .1, .2)
• Enhanced _id_like_percentage function with configurable seed parameter for reproducible random
 sampling
• Increased sample size from 500 to 2000 for better statistical representation
• Implemented conditional random sampling when values exceed sample size threshold

cytetype/preprocessing/validation.py

---

[qodo-code-review]

Code Review by Qodo

🐞 Bugs (2) 📘 Rule violations (0) 📎 Requirement gaps (0)

1. Warning example mismatch 🐞 Bug ✧ Quality

Description

_id_like_percentage now computes pct from a random sample, but warnings still show example IDs from
the first 20 entries. This can produce confusing messages (e.g., pct>50 with empty/unrepresentative
examples).

Code

cytetype/preprocessing/validation.py[R80-89]

+def _id_like_percentage(values: list[str], seed: int = 42) -> float:
    if not values:
        return 100.0
-    n = min(500, len(values))
-    sample = values[:n]
+    n = min(2000, len(values))
+    if n < len(values):
+        rng = random.Random(seed)
+        sample = rng.sample(values, n)
+    else:
+        sample = values
    return sum(1 for v in sample if _is_gene_id_like(v)) / n * 100

Evidence

The percent is computed from rng.sample(values, n) (random subset), while the warning examples are
derived from values_list[:20] (a different, deterministic subset), so the warning can cite
examples that don’t correspond to the sampled population that produced pct.

cytetype/preprocessing/validation.py[80-89]
cytetype/preprocessing/validation.py[103-110]

Agent prompt

The issue below was found during a code review. Follow the provided context and guidance below and implement a solution

### Issue description
`pct` is computed from a sampled subset of values, but `example_ids` comes from the first 20 values. This can produce warnings where the % suggests many IDs but the examples list is empty or misleading.

### Issue Context
The mismatch was introduced when `_id_like_percentage` switched from `values[:n]` to random sampling. The warning content should stay consistent with how `pct` was computed.

### Fix Focus Areas
- cytetype/preprocessing/validation.py[80-113]

### Implementation notes
- Option A (best): Change `_id_like_percentage` to optionally return both `(pct, sample)` so callers can reuse the exact sample.
- Option B (localized): In `_validate_gene_symbols_column`, compute `example_ids` by scanning `values_list` (not just first 20) until 3 ID-like values are found.
- Option C: In `_validate_gene_symbols_column`, mirror `_id_like_percentage` sampling logic locally so `pct` and `example_ids` use the same `sample` list.

ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools

---

2. Higher validation CPU cost 🐞 Bug ➹ Performance

Description

_id_like_percentage now checks up to 2000 values instead of 500 and is called repeatedly during
gene-symbol auto-detection across many columns. This increases total regex work and could slow
validation on datasets with many candidate columns.

Code

cytetype/preprocessing/validation.py[R83-86]

+    n = min(2000, len(values))
+    if n < len(values):
+        rng = random.Random(seed)
+        sample = rng.sample(values, n)

Evidence

The per-call cap increased to 2000, and the function is invoked in multiple loops over adata.var
columns (known columns, var_names, and then all other columns), multiplying the added per-call cost
by the number of candidate columns.

cytetype/preprocessing/validation.py[80-89]
cytetype/preprocessing/validation.py[157-202]

Agent prompt

The issue below was found during a code review. Follow the provided context and guidance below and implement a solution

### Issue description
`_id_like_percentage` increased its sample cap (now up to 2000). Because `resolve_gene_symbols_column` calls it for many candidate columns, this raises total regex work and may slow validation for unusually wide `adata.var` tables.

### Issue Context
This is a trade-off: higher `n` reduces sampling variance but increases CPU time. The best fix depends on typical dataset sizes in your users’ workflows.

### Fix Focus Areas
- cytetype/preprocessing/validation.py[80-90]
- cytetype/preprocessing/validation.py[145-202]

### Implementation notes
- Add a module-level constant (e.g., `ID_LIKE_SAMPLE_CAP = 500|1000|2000`) or make it a parameter passed from `resolve_gene_symbols_column`.
- Precompile the regexes used by `_is_gene_id_like` (if profiling shows regex dominates).
- Optionally lower the cap when iterating many columns (e.g., `cap = min(2000, max(200, 2000 // n_candidate_columns))`).

ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools

---

ⓘ The new review experience is currently in Beta. Learn more