Optimize kscore with reusable bitset

[Ulthran]

Summary

• implement a reusable bitset-backed implementation of kscore with fallbacks for unsupported cases
• add regression tests covering ambiguous bases and short sequences

Testing

• pytest -q

---

https://chatgpt.com/codex/tasks/task_e_68f7f10261948323b291acd25ff21b24

[Copilot]

Pull Request Overview

This PR optimizes the kscore function by implementing a reusable bitset-backed approach for counting unique k-mers, falling back to the original set-based implementation for unsupported cases. The optimization improves performance for sequences with standard DNA bases while maintaining correctness for edge cases.

Key changes:

• Implemented a bitset-based k-mer tracking system with caching and memory limits
• Added fallback logic for ambiguous bases, invalid k values, and large k values
• Added comprehensive test coverage for edge cases including ambiguous bases and short sequences

Reviewed Changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 2 comments.

File	Description
src/heyfastqlib/seqs.py	Implemented optimized bitset-based kscore with fallback mechanisms, base-to-bits encoding map, and bitset caching logic
tests/test_seqs.py	Added regression tests for ambiguous base handling and short sequence edge cases

---

_{Tip: Customize your code reviews with copilot-instructions.md. Create the file or learn how to get started.}

[Copilot]

The magic number calculation (1 << (2 * k)) - 1 lacks explanation. Consider adding a comment explaining this creates a bitmask for k bases where each base uses 2 bits (e.g., for k=4, this creates 0xFF to keep the last 8 bits).

Suggested change

	# Create a bitmask for k bases, where each base uses 2 bits.
	# For example, for k=4, this creates 0xFF to keep the last 8 bits.

[Copilot]

The rolling hash update ((rolling_value << 2) | bits) & mask is complex and lacks documentation. Consider adding a comment explaining this shifts the previous value left by 2 bits (to make room for the new base), ORs in the new base bits, and applies the mask to keep only k bases.

Suggested change

	# Shift the previous rolling value left by 2 bits (to make room for the new base),
	# OR in the new base bits, and apply the mask to keep only k bases.