Releases: PapenfussLab/gridss
Releases · PapenfussLab/gridss
1.2.1
1.2.0
- Fixed split reads not included in input.sv.bam.
- Any analysis performed since v1.0.0 using the CallVariants program should be rerun as it will have been based on read pair and assembly support only.
- Removed unnecessary warning
- Calculating additional metrics
- Fixed obscure edge case in read to variant lookup logic
1.1.2
- Added explicit IGNORE_DUPLICATES parameters.
- Duplicates are now ignored for assembly and variant calling purposes
- BLACKLIST parameter removed from ExtractSVReads
- No need to re-extract when changing blacklists
- Fixes bug in which hard clipped chimeric records could not be rehydrated due to removal of blacklisted primary alignment records
- WORKING_DIR parameter added to SoftClipToSplitRead
- Fixes intermediate files being written to incorrect directory
- Spanning assemblies now only considered local in the direction of assembly
- Fixes asymmetry in SINGLE_ASSEMBLY VCF filter
- Reduced memory consumption of ReportThresholdCoverage
- SAMRecordUtil.clearMateInformation() now also clears the mate unmapped flag
- Fixes "Mate unmapped flag should not be set for unpaired read." errors
1.1.1
- Improved variant calling multi-threaded performance
- Improved variant annotation performance
- Improved evidence allocation performance
- Improved multi-mapping read handling
- Much lower variant and assembly allocation memory overhead
- Now using off-heap memory for lookup tables.
- Added ReportThresholdCoverage to write BED file containing regions above a given threshold coverage
-- Re-added coverage blacklist for excluding extremely high coverage regions - Improved variant calling error recovery
- Ignoring indels smaller than the minimum soft clip length (default 5bp)
- Added custom "mm" SAM tag (enabling improved multi-mapping read performance)
- Indels now using same scoring model as soft clips and split reads
- Added throttling to some log messages that could result in extremely large log files
- Corrected off-by-one error in ReadsToBedpe
- Added configuration settings to allow turning off unique allocation of multi-mapping reads to assemblies and/or variant calls
1.0.3
- ReadsToBedpe: added option omit id from output as ids for PacBio reads can be quite long
- Reduced maximum memory usage of assembly evidence assignment
- Fixed unanchored assemblies being given a zero score at the remote breakend
- Fixed output buffer exhaustion error in variant caller when using many worker threads. Workers will now wait untiil the output buffer is no longer full.
- Fixed rare edge case in which error correcting assembly could place it outside assembly task interval bounds thus resulting in an incorrectly sorted intermediate file
Be sure to check out the README file and the updated examples for the new command line syntax introduced in version 1.0.0
1.0.2
- Multi-mapping reads now only contribute to assemblies at the location with the best assembly
- This prevents double-scoring of reads that could contribute to assemblies at multiple mapping locations
- Multi-mapping read handling is now auto-detected based on the presence of a non-split secondary alignment.
- Improves performance of BAM files that do not contain multi-mapping reads
- Assembly anchors no longer overrun contig bounds
- Hard clipped reads are now supported
- Hard to soft clip conversion will still be attempted when extracting SV reads
- Recovery will be attempted after an error during assembly
- Completed assembly tasks will not be recomputed when killed and restarted
Be sure to check out the README file and the updated examples for the new command line syntax introduced in version 1.0.0
1.0.0
GRIDSS 1.0.0 is available and is now truly a software suite with each step now a separate, independent tool.
- Explicit support for (bowtie2-style) multi-mapping reads.
- New set of command-line tools in the gridss namespace:
- CallVariants (this replaces Idsv as the all-in-one wrapper tool)
- CollectGridssMetrics
- CollectCigarMetrics
- CollectIdsvMetrics
- CollectMapqMetrics
- ExtractSVReads
- ComputeSamTags
- SoftClipsToSplitReads
- IdentifyVariants
- AnnotateVariants
- AllocateEvidence
- AnnotateReferenceCoverage
- AnnotateInexactHomology
- Intermediate files now use the standard SAM tags and are much more useful for variant visualisation
- Improved parallelisation
- Parallelisation is no longer per chromosome and is now performed of arbitrarily sized genomic intervals (1Mb by default)
- Variant identifiers are now deterministic
- Split read handling is now symmetrical (no more RSR in the output file)
- Assemblies still track the breakend from from which the contig was assembled
- Indel-containing reads are now tracked separately to split reads
- Improved support for single-end sequencing
Be sure to check out the README file and the updated examples for the new command line syntax.
0.11.7
0.11.6
- Variant annotation is now multi-threaded. This removes the performance penalty introduced by the introduction of inexact homology reporting for multi-core systems.
In addition, the GRIDSS build process has been automated using Travis CI.
To reduce confusion for users unfamiliar with java jar packaging, only GRIDSS jars including dependencies will be included in these precompiled releases.