Releases: AstraZeneca-NGS/VarDictJava
Fixed end coordinates for insertions
Fixed end coordinates for insertions (minor issue)
More Chimeric improvements
More chimeric improvements
Java 10 compatibility
Gradle update
Chimeric and unique modes
Merge pull request #108 from PolinaBevad/unique_mode_by_read_number Added unique mode by read number (option -UN)
VarDict 1.5.1: Updating submodule
VarDict 1.5.0: Fix for multi-allelic site and handling of leading and trailing insertions in cigar strings
Fix for Overlapping reads problem (updated)
Fix for Overlapping reads problem (updated)
VarDict-1.4.10.zip
VarDict 1.4.9: fix for Overlapping reads
VarDict 1.4.9: fix for Overlapping reads
VarDict-1.4.9.zip
VarDictJava refactored (2.0) draft
Fix for excessive reference file openings, updated htsdk version to 2.8
Fixed problem with excessive reference file openings. Used the same approach as with SamReader.
Also updated htsjdk version to 2.8 and JDK to 1.8.
Release 1.4.7: bugfix, performance and resource usage improvement
Fixed problem revealed in issue #56 (VarDict Amplicon Mode throws IndexOutOfBoundsException). Solution: Mapped reads are skipped now (See ca0acbb )
Performance and Resource usage Improvement.
VarDict created new SAMReader for each interval from input .bed file. Recreating the same readers treating file system thousands of times is not supposed way of using SamReaderFactory. In some environment it might cause problems due to heavy file system interaction. Anyway, multiple recreating of the same object is not efficient in terms of performance.
We cache Readers for each thread of VarDict now. It leads to the restriction that the only SamView object must be created per thread (and only after the previous one is closed). The is no such a case in VarDict.
Apart from resolving file system interaction problems, Readers caching shows better resources utilization and speeds up overall work. Our tests show up to 30% speedup on some cases (the best results are shown in single-threaded mode).