Releases · divonlan/genozip

Reference files cached in shared memory are now removed after 24 hours of non-use
Support for "minimal" references (see genozip.com/reference)
diagnostics: --show-seg-summary

15.0.81 3/4/2026

Improved references files for better FASTQ compression. With reference files created
(genozip --make-reference) with this release:

FASTQs will compress ~20-30% faster, and about 3-5% better
the reference file now supports references up to 1 Tbps (1 trillion bases).
Existing (older) reference files remain supported as well.
RAM consumption during FASTQ compression is reduced by 800 MB
Switched default integrity verification from Adler32 to better XXH3. no change in MD5 availablability with --md5.
Added support for FASTQs with empty reads (seq_len=0)
Resolved activation issues in China
diagnostics: --submits-stats renamed to --telemetry, --debug-progress renamed --show-tasks, --debug-submit cancelled

Significant reduction in RAM consumption (up to 33% in some cases) in genounzip (inc. in the test phase at the end of compression)
Significantly faster compression of many Illumina FASTQ files due to better parallelization (IL4M)
Mac (Apple processor) now released via conda too.
New: 'genozip --is-exactable' inspects the gz-compression of a file and reports whether genounzip would be able to gz-recompress the file to the same exact gz-compression if --bgzf=exact is used.
New: 'genounzip --bgzf=exact-strict' option: same as --bgzf=exact, but fails if gz-recompression cannot be done exactly, instead of falling back to the default re-compression.
New diagnostic options: --show-flavor, --dump-gz-block. See genozip.com/diagnostics

--index is now default for genounzip and genocat for bam, sam.gz and vcf files, and the index files are generated natively without any performance overhead.
--no-bai and --no-tbi and new genounzip/genocat options for overriding the default --index
genozip --skip-index is a new options - skips compressing bai, tbi, csi, crai and gzi files.
genocat --converage and --idxstats now also works on some FASTA files (those that are essentially QUAL-less FASTQs)
New diagnostic option 'genozip --show-bai file.bai' and --show-tbi to see content of .bai/.tbi files. See: genozip.com/diagnostics.
New diagnostic option --debug-bai
Canceled diagnostic option --no-zriter

FASTA: support --bamass, --deep and --optimize as in FASTQ (but not --pair yet),
Deep: new genocat --fasta option for displaying the fasta component of a Deep file with FASTA
New genocat option for SAM/BAM: --qname-only
If GZ block identification is wrong, restart with --no-bgzf
Minor bug fixes

Discontinue support for Mac with Intel processors (now only ARM / Mx processors are supported for Mac)
FASTQ with --bamass: compression improvements for MGI data
Small compression improvement for minimap2 BAMs
Don't fail for SAM/BAM files with invalid QUAL score values
--bamass now ignores consensus reads contained in a BAM file
genocat --print-reference can now be combined with --md5 to show the MD5 only (see genozip.com/genocat)
--echo can now receive an argument - useful for adding contextual information when the logging output
new diagnostic --show-TLEN-pred (see www.genozip.com/diagnostics)

Releases: divonlan/genozip