Hello CLARK users,
A new CLARK release is now available: Version 1.4.5.0-a, released on June 21, 2026.
This release continues the recent work to modernize and harden CLARK’s RefSeq database setup workflow, with a particular focus on large RefSeq downloads, resumability, correctness checks, and better reporting.
The main update in this release is that the RefSeq sequence-transfer loop has been replaced with a Python 3 standard-library downloader. This removes reliance on the previous shell-based transfer path and provides a stronger foundation for reliable, maintainable downloads.
Key improvements include (cf. CHANGELOG):
- Bounded parallel RefSeq downloads
- Resumable .part files for interrupted transfers
- Gzip validation for downloaded files
- Deferred retry handling for transient failures
- Aggregate progress reporting
- Clear failure reports
- Regression test coverage for successful downloads, transient failures, deferred retries, permanent failures, resume behavior, and supported database categories
This release preserves the existing RefSeq selection and provenance behavior for bacteria, viruses, plasmid, plastid, protozoa, and related databases.
Recent versions also introduced several important improvements leading up to this release, including:
- Improved RefSeq database setup for large downloads, especially bacteria
- Parallel resumable RefSeq transfers by default
- Optional filters for smaller exploratory databases
- NCBI URL validation, retries, gzip validation, partial-file cleanup, and aggregate progress reporting
- Use of RefSeq assembly-summary provenance to avoid large global accession-map lookups when taxids are already available
- Simplified CLARK classifier dispatch and stricter --kso mode validation
- Additional unit and regression tests for file helpers, HashTop, analysis tools, abundance handling, sequence extraction, summary tables, and target k-mer distribution edge cases
- RefSeq downloader dry-run support, manifest output, and provenance output
- Repository and script reorganization
- Build workflow update: users should now build CLARK with make all
Users are encouraged to update to the latest release, especially if they work with large RefSeq databases or have previously encountered interrupted or unreliable database downloads.
Thank you for using CLARK.
Best,
Rachid