1 Commits

Author SHA1 Message Date
Eric Coissac
f78543ee75 Refactor k-mer index building to use disk-based KmerSetGroupBuilder
Refactor k-mer index building to use the new disk-based KmerSetGroupBuilder instead of the old KmerSet and FrequencyFilter approaches. This change introduces a more efficient and scalable approach to building k-mer indices by using partitioned disk storage with streaming operations.

- Replace BuildKmerIndex and BuildFrequencyFilterIndex with KmerSetGroupBuilder
- Add support for frequency filtering via WithMinFrequency option
- Remove deprecated k-mer set persistence methods
- Update CLI to use new builder approach
- Add new disk-based k-mer operations (union, intersect, difference, quorum)
- Introduce KDI (K-mer Delta Index) file format for efficient storage
- Add K-way merge operations for combining sorted k-mer streams
- Update documentation and examples to reflect new API

This refactoring provides better memory usage, faster operations on large datasets, and more flexible k-mer set operations.
2026-02-10 06:49:31 +01:00