0.6 (05/12/17)

Added splitting script to allow parallelization
Testing now uses custom script instead of diff, allowing for small differences resulting from differing Perl version
Removed -h
Removed setup.sh
Added -o option to generateannotations.pl
Added -t to control memory usage
Simplified test.sh, made output look prettier, now works from any directory
Added compatibility with newer versions of svtools, along with -s option for nonstandard svtools installations

0.5.1 (06/20/16)

Calculate breakend scores based on scores at possible breakends, not bases
Revamped conditions for calculating truncation scores
Rescale probability distributions based on which possible breakpoints have available scores
Rescale probability distribution again for TOP[n] operations
Added stress test
All scores with weighted operations for variants with no PRPOS are now -1
Changed usage (-i specifies input file, -o now takes comma-delimited list of operations, removed -s)
Added script which downloads/generates annotation files from single annotation track
Added test for PRPOS in header if a weighted operation is specified - defaults to unweighted if not found
Removed -w, added TOP[n]WEIGHTED and MEANWEIGHTED as operations
Default operation is now TOP10WEIGHTED
Added INS support, removed TRX support
Allow for piping VCF files through standard input
Use exact breakpoint representation of SVs in BEDPE (in conjunction with new version of svtools)
Calculate span scores between most likely bases
Calculate truncation scores for DELs
Treat MEIs like BNDs
Calculate truncation scores from most likely bases
Added setup script
Bug fixes/code improvements
Error handling

Account for regions of the genome which do not have CADD scores
Use vcfanno 0.0.8
Fixed sorting incompatibilities with vcfanno
Implemented -o TOP (mean of top n bases) and -o MEAN options
Added -t (number of bases to use in -o top), -p (gene file strand column), and -v (verbose) options
Removed exon annotation and -n (SVScore still requires an exon file to create intron file)
Added support for CNV and MEI variant types (treated as DELs/DUPs)
Replaced BND with TRX, BNDs still supported but only left/right breakend scores are calculated
Updated added header lines
Temporary files go in svscoretmp directory, which is deleted automatically (except under -d)
Made introns.bed filename more descriptive to avoid problems in switching between annotation systems in the same directory
Improved performance by returning a score of -1 for regions which are too big (above 1 Mb)
Added unit test
Allow for calculation of scores for variants reclassified to BND without changing the ALT field

Performance improvement (fixed high memory usage bug)
Use vcfanno v0.0.7 and CADD v1.3
Fixed off-by-one error in coordinates
Added -g, -e, -m, and -n options to allow use of custom gene/exon annotation files
Processes BND variants without mates, provided CIEND field is present
Updates VCF header with scores added to INFO field
Intermediate files are tagged with timestamp to prevent collisions from parallel instances
Automatically detects whether file is gzipped based on filename (removed -z option)
Always calls vcfanno (removed -a option)

Outputs sorted VCF to standard output, no longer creates *.scored.txt file
Relaxed stringency on SVTYPE field
File prefix now disregards path prefix to always work in current directory
Removed natural sort requirement. SVScore now automatically sorts input VCF based on ID
Returns a negative score for variants in regions with no C scores (e.g. GL contigs)
Returns a score of 100 for deletions/duplications above 1 Mbp