FLAIR Release 3.0.0

[v3.0.0] 2025-11-31

• General
  • Bug fixes since v3.0.0b1
  • Removed dependencies on pandas and rpy2.
  • Added proper logging throughout
• FLAIR align
  • Cleaned up output files - now outputs unfiltered BAM and filtered BED
• FLAIR correct
  • Changed orthogonal junction inputs - now specify whether your file is
    a bed (--junction_bed) or whether it comes from STAR short-read RNA alignment (--junction_tab)
  • Now support detecting junctions directly from long read data:
    run intronProspector to generate junction bed, input that via --junction_bed,
    specify desired junction read support with --junction_support
  • Removed -g genome option, increases speed
  • Now produces sorted output files
• FLAIR collapse
  • cleaned up output files, now only gives: isoforms.bed, isoforms.gtf, isoforms.fa, isoform.read.map.txt
  • (experimental) added filter for removing reads with internal priming.
    Options: --remove_internal_priming, --intprimingthreshold, --intprimingfracAs
  • allow CDS prediction in collapse with --predictCDS
  • removed genomic range option (too fragile). For parallelization, run FLAIR transcriptome
  • made gtf (annotation) parsing more robust, especially for unsorted files
  • improved fractional support filtering (with --support < 1)
  • improved isoform haplotyping through longshot - this is being deprecated though, please use FLAIR variants
• FLAIR quantify
  • improved fraction of reads assigned to isoforms and recovered (better handling of ambiguous alignments)
  • improved processing of multiple samples so running requires less available memory and is faster
• FLAIR diffexp and diffsplice
  • recoded directly in R to improve performance
• New modules
  • FLAIR transcriptome
    • Combines the functions of correct and collapse
    • Runs directly from an aligned BAM file
    • Performs more effective parallelization (specify in --parallelmode)
  • FLAIR fusion
    • Detects gene fusions and fusion isoforms
    • Fusion detection accuracy is comparable to JAFFAL and CTAT-lr-fusion
    • Fusion isoform detection is much more accurate
  • FLAIR combine
    • Allows combining transcriptomes generated from different samples
    • Fusion isoforms can be combined with collapsed isoforms to form a full transcriptome
    • Manipulate filtering of single exon isoforms with --include_se
  • FLAIR variants
    • Allows detection of variant-aware transcripts through
      identification of read clusters with shared variants
    • Uses variants detected either from WGS or from lr-RNA-seq.
      We recommend Longshot.
    • Good for identifying splicing of specific variants
    • Can be used in conjunction with FLAIR diffexp or diff_iso_usage
      to identify changes in variant-aware transcripts between samples/groups
• Added FLAIR protocol to improve guidance on running flair

FLAIR v3.0.0 beta 1

FLAIR v3.0.0 beta 1 release

Major user-visible changes in v3.0.0b1

• General
  • Removed dependencies on pandas and rpy2.
  • Added proper logging throughout
• FLAIR align
  • Cleaned up output files - now outputs unfiltered bam and filtered bed
• FLAIR correct
  • Changed orthogonal junction inputs - now specify whether your file is
    a bed (--junction_bed) or whether it comes from STAR short-read RNA alignment (--junction_tab)
  • Now support detecting junctions directly from long read data:
    run intronProspector to generate junction bed, input that via --junction_bed,
    specify desired junction read support with --junction_support
  • Removed -g genome option, increases speed
  • Now produces sorted output files
• FLAIR collapse
  • cleaned up output files, now only gives: isoforms.bed, isoforms.gtf, isoforms.fa, isoform.read.map.txt
  • (experimental) added filter for removing reads with internal priming.
    Options: --remove_internal_priming, --intprimingthreshold, --intprimingfracAs
  • allow CDS prediction in collapse with --predictCDS
  • removed genomic range option (too fragile). For parallelization, run FLAIR transcriptome
  • made gtf (annotation) parsing more robust, especially for unsorted files
  • improved fractional support filtering (with --support < 1)
  • improved isoform haplotyping through longshot - this is being deprecated though, please use FLAIR variants
• FLAIR quantify
  • improved fraction of reads assigned to isoforms and recovered (better handling of ambiguous alignments)
  • improved processing of multiple samples so running requires less available memory and is faster
• FLAIR diffexp and diffsplice
  • recoded directly in R to improve performance
• New modules
  • FLAIR transcriptome
    • Combines the functions of correct and collapse
    • Runs directly from an aligned BAM file
    • Performs more effective parallelization (specify in --parallelmode)
  • FLAIR fusion
    • Detects gene fusions and fusion isoforms
    • Fusion detection accuracy is comparable to JAFFAL and CTAT-lr-fusion
    • Fusion isoform detection is much more accurate
  • FLAIR combine
    • Allows combining transcriptomes generated from different samples
    • Fusion isoforms can be combined with collapsed isoforms to form a full transcriptome
    • Manipulate filtering of single exon isoforms with --include_se
  • FLAIR variants
    • Allows detection of variant-aware transcripts through
      identification of read clusters with shared variants
    • Uses variants detected either from WGS or from lr-RNA-seq.
      We recommend Longshot.
    • Good for identifying splicing of specific variants
    • Can be used in conjunction with FLAIR diffexp or diff_iso_usage
      to identify changes in variant-aware transcripts between samples/groups
• Added FLAIR protocol to improve guidance on running flair

Release 2.2.0

[2.2.0] 2025-05-06

• Returned diffexp and diffsplice as standard modules. The BioConda
  environment does not include the dependencies for these modules
  and required software does not run on Apple Silicon (ARM64) systems.
• The flair combine functionality is now to a module. The
  flair_combine program is run with flair combine.
• Changed default MAPQ minimum quality score to 0. This allows more reads to
  be used in identifying isoforms, which tends to improve the overall models
  with out adversely affecting the accuracy.
• GitHub releases include the Conda YAML files for building FLAIR
  environments. Useful if the BioConda release has not been manually
  reviewed.
• The FLAIR Docker now includes all dependencies to run diffexp and diffsplice.
• Reorganized the installation documentation.
• Fixed flair_quantify --output_bam crash
• Other bug fixes

Release 2.1.2

[2.1.2] 2025-04-17

• Address issue getting BioConda to work
• Bug fixes for collapse command line parsing

FLAIR 2.1.1

[2.1.1] 2025-04-10

• converted all programs to use console scripts to allow BioConda to work

FLAIR 2.1.0

[2.1.0] 2025-03-27

• Numerous bug fixes.
• Removed support for PSL format.
• Remove flair 123 to run multiple modules at once.
• Compatibility with Python 3.12
• Compatibility with Apple ARM64 systems.
• Deprecated diffExp and diffSplice, they will be removed in a future release.
  Lets us know if you use this functionality. Their dependencies are no longer
  part of the conda package, they can be added the conda environment with
  misc/flair_diffexp_conda_env.yaml.

Flair 2.0

This release accompanies the Flair 2.0 paper https://www.biorxiv.org/content/10.1101/2023.06.09.544396v1

• Fixes tickets (#261 #234)
• Moves or links some scripts to /bin so users have access
• Adds FAQ; updates documentation
• Removes sam flag name extension from bed files because users have trouble getting it to parse their own IDs properly.
• Adds --nvrna tag to flair quantify to allow for strand-aware alignment, like used in flair align
• Adds quoting to handle wildcard characters in filenames
• Removes obsolete fusion_dist option from collapse

Flair 1.7

• adding user tests
  

• added cannot_verify output file to ssCorrect to avoid flair correct silently removing reads for which the chromosome cannot be found in the input annotation
  

• removed psl from main Flair to simplify code and inputs. Conversion tools remain in bin directory. Programs in src directory still take psl but are renamed to avoid confusion
  

• removed file path inputs to flair module parameters to simplify code and input. With conda or docker install these are never necessary
  

• full usage statements for each flair module in documentation
  

• info on testing local installation in documentation
  

• removed salmon option
  

• standardized es output with other diffSplice output (added gene ID) and removed strand from ir output
  

Changes to diffExp, diffSplice output and error/warning messages

• Reorganized diffSplice and diffExp outputs; added human readable p-adj filtered tables
• Improved diffSplice error and warning notifications
• flake 8 code cleanup
• Fixed wiggleWindow bug (was not getting passed on from flair to ssPrep) and set default to 15
• Reorganized usage statements
• Added sanity checks and warnings

Fixed reliance on specific bedtools version

This version no longer specifically needs bedtools v2.25
There is also a new docker container on dockerhub: brookslab/flair:1.6.3