v0.2.0

CRISPR-HAWK v0.2.0 – Release Notes

Date: April 28, 2026

This major update transforms the framework into a comprehensive scoring powerhouse by integrating three state-of-the-art efficiency models and introducing an automated environment and model management system.

What's Changed

• Advanced Scoring Integration: We have expanded our predictive capabilities by integrating sgDesigner, PLM-CRISPR, and CRISPRon. These models provide high-confidence efficiency metrics for your guide designs. (#8)

• Automated Model Catalogue: No more manual downloads. CRISPR-HAWK now automatically detects missing model weights, downloads them with built-in retry logic, and manages the extraction process. (#10)

• Intelligent Environment Management: Using the new ScoringEnvs system, the tool now handles complex dependencies (like Perl/Python wrappers for sgDesigner) via automated Conda/Mamba execution. (#10)

• Codebase Hardening: We've streamlined the internal API, enforced strict sequence padding for deep-learning compatibility, and removed legacy modules (bitset.py, microhomology.py) to improve maintainability. (#10)

Backwards Compatibility / Migration Notes

• Output Format: Three new columns (score_sgdesigner, score_plmcrispr, and score_crispron) have been added to the output TSV. This shifts the index positions of subsequent columns. If you use downstream scripts, please switch from hardcoded column indices to header-name parsing. (#10)

• Sequence Padding: A strict GUIDESEQPAD = 10 is now enforced in the Guide class to satisfy the input requirements of the new deep-learning models. (#8)

New Contributors

• @alisakumbara

Changelog

Full Changelog: v0.1.2...v0.2.0