Releases · GoekeLab/bambu

16 Jun 01:22

cying111

Latest

Changes since last release v2.0.0

Major changes:

Updated the input parameters of Bambu to simplify the user experience
Introduced NDR threshold recommendation
Implemented trainBambu(), allowing users to train and use models on their own data
Reads that cannot be assigned to any transcript are grouped as incompatible counts
Partial estimates are removed from output as it can be directly obtained based on total count estimates and full-length count estimates
The fusion mode is now available, which assigns read classes that align to multiple genes to a new combined fusion gene

Minor changes:

Novel transcripts and genes are now by default output with a Bambu prefix
Updated the documentation, messages and errors output by Bambu
Annotated transcripts (with unique exon-junctions) with at least 1 full-length read are assigned a NDR rank
Fixes crash during Low Memory Mode when there are scaffolds with no reads
Fixes crash on windows machines caused by DNAStringSet
Adds NDR metadata when running discovery mode with recommended NDR, so users do not need to look at console for the recommended NDR.
Re-enabled GitHub actions for new devel branch name and the windows check
Fixed a crash that occurs with large datasets resulting in large overflow tables during novel gene id assignment
Remove nested bplapply in EM
Remove unused eqClassById list column in the readClassDist object to reduce memory usage
Fixed a bug that caused identical unspliced reads to not be tracked when trackReads = TRUE

Assets 2

03 Nov 05:32

cying111

bambu v2.0.0-Bioconductor release

Major Changes:

Implemented a machine learning model to estimate transcript-level novel discovery rate
Implemented full length estimates, partial length estimates and unique read counts in final output
Improved the performance when extending annotations with simplified code
Improved the performance when large amounts of annotations are missing.
Implemented a lowMemory option to reduce the memory requirements for very large samples (>100 million reads)

Minor fixes:

remove the use of get() which looks into environment variables (prone to crashes if a variable of the same name exists) and directly references the functions that should be used instead.
bug fix when a fa file is provdied as a string variable to non-windows system
bug fix when no single exon read class in provided samples
bug fix when no splice overlaps found between read class and annotations

Assets 2