This release fixes mostly minor bugs and improves scalability by reducing memory and facilitating parallelization.

What's Changed

• Reduce memory by @brentp in https://github.com/quinlan-lab/STRling/pull/109
• Merge by chrom workflow by @hdashnow in https://github.com/quinlan-lab/STRling/pull/111
• zero length reads by @christopher-schroeder in https://github.com/quinlan-lab/STRling/pull/106
• report 0-len alignments in debug mode. by @hdashnow in https://github.com/quinlan-lab/STRling/pull/112
• add empty line for code block rendering on github by @kew24 in https://github.com/quinlan-lab/STRling/pull/115

New Contributors

• @christopher-schroeder made their first contribution in https://github.com/quinlan-lab/STRling/pull/106
• @kew24 made their first contribution in https://github.com/quinlan-lab/STRling/pull/115

Full Changelog: https://github.com/quinlan-lab/STRling/compare/v0.5.1...v0.5.2

Updates to documentation only. Created an extra release for the STRling paper.

Code updates: Let outliers script take a glob file pattern (useful when number of input files is extremely large) Merge checks that input file exists

Updates to docs

• Add STRling logo
• Add workflows
• Release instructions

Fix travis CI

Improved accuracy of STR locus bounds:

• for bounds calc, ignore soft-clips that are inconsistent with the center mass
• set min-support default to 5

More robust statistics in the outlier test

• Switch to using hubers estimates of mu/sd in outlier test
• Give huber 1000 iterations, then fallback to MAD (optionally report which was used)

Documentation:

• Add instructions for creating a release and static binary

Pipeline fix:

• Rename outliers -> outliers_py to avoid name conflict

• Introduces conda install instructions in the docs
• Make the merge command less verbose by default
• Make warnings/errors more useful and correct

this release is unchanged from 0.4.0 except that it tightens a dependency in nimble which has been updated.

Major:

• Outlier testing (Python script)
• Install with conda
• Updated default read support thresholds and more internal consistency with their use

Minor:

• Allow mixing of samples aligned to different reference genomes in merge (off by default, use with caution)
• Fix Travis CI testing
• Docs updates

Major:

• When doing joint calling, merge subcommand only reports bounds with a minimum number of supporting reads in at least one sample (default: 6)
• Introduces index subcommand to create an str index of the reference genome. This was previously done per sample, but is more efficient to do once in advance for larger cohorts

Minor:

• Memory required for merge reduced by more than half
• Fix int overflow bug
• Docs updates

• Fixes "SIGSEGV: Illegal storage access" error when parsing bed files
• Informative error when bed file includes a repeat unit > 6 bp

Major

• Reduction in false positive calls via changes to the way repeat units are counted in whole and soft-clipped reads
• Calculate expected vs observed number of spanning fragments
• Add command to extract informative reads for a locus from bam/cram

Minor

• Force reporting/genotyping of loci provided in an input bed file
• Provide STR disease loci for hg38
• End-to-end testing for CI
• Updated documentation