IRMA v1.3.0

Release Notes for v1.3.0 : 2025-06

Added

• Created a Dockerfile for building Linux x86_64 and aarch64 containers
• Added Github actions for generating containers and packaged binaries on release

Changes

• IRMA has been relicensed under Apache 2. We have more clearly separated third
  party software that is distributed with IRMA under other licenses and
  conditions. See <IRMA_RES/third_party/MANIFEST.md> for more details.
• LABEL v0.7.0+ is now required to run IRMA.
• IRMA-core is now required to run IRMA and is packaged with it from now on.
  Certain read preprocessing and merging steps are now handled by IRMA-core.
• IRMA now provides automatic core detection for local execution mode.
  SINGLE_LOCAL_PROC can be still be used up to the available cores.
  LOCAL_PROCS_OVERRIDE can be set to pick any number of cores.
• Third-party binaries have been standardized to newer versions and generally
  support Linux, macOS on both aarch64 and x86_64. Use of LABEL classification
  on macOS still requires Rosetta.
• IRMA's updated, third-party SSW (v1.2.5M for modified) now provides soft
  clipping. This can slightly increase the size of intermediate data.
• Documentation has been updated for Github
• IRMA's QC_log.txt has been completely re-formatted and updated.

Fixes

• IRMA no longer uses which, which could sometimes cause warnings in some
  environments.
• The previously pinned version of SSW had a bug that shifted quality scores in
  reverse-complemented reads that were also clipped by the alignment (<8%
  observed). The newest version fixes this bug and another rare alignment
  correctness error (<0.01%).

Deprecations and Removals

• DOUBLE_LOCAL_PROC has been removed and will only issue a warning if set.
• USE_IRMA_CORE is deprecated and will be removed in a future version.

New Known Issues

• Some users have noted H1 swine misclassification via BLAT (LABEL can be used
  as a workaround). We will address this with a reference seed update.
• Merged-pair BAM files do not show clipping information. We will add clipping
  information for merged-pair reads in a future release.

v1.2.1 : 2024-10

Fixes

• The S-gene only SC2 configuration now has consensus padding turned off. The module is not compatible with this feature.
• IRMA had reported reads passing only the length filter as the observed raw reads. This has been fixed.