Trans-ABySS 1.3.5 (Sep 13, 2012)
Lowered contig size filtering cutoff from 2k-1 to 2k-10; various improvement to fusions and indel code (see ChangeLog)
For Linux (1.3 MB)
|License||BCCA (academic use)|
|Release Manager||Readman Chiu|
The main change for v1.3.5 is that the minimum contig size cutoff has been lowered from 2k-1 to 2k-10 in order to capture more junction contigs. The major changes to the analysis modules since the previous public release (v1.3.2) are documented in the ChangeLog.
# lower the contigs filtering threshold from 2k-1 to 2k-10
# "abyss-ta-filter" and "abyss-rmdups-iterative" updates: can work with either "contigs and indel-bubbles" or "indels only"
# fusion bug fix: only label contig as 'multi-mapping' (and hence screening it out) only if all more than 1 possible pairs pass read support
# get rid of reliance on ANCHOR to align reads to contigs
# use of breakpoint pairs (breakpoint pairs) for read support. Breakpoint pair: 1 mate spans the event breakpoint (hence only aligned in reads-to-contigs alignmen
ts) and the other mate mapped on either side of the breakpoint and pointing
# original 'read_pairs' becomes 'flanking_pairs'
# final number of read pairs used for filtering = num_flanking_pairs + minimum of the 2 num_breakpoint_pairs
# if fusion transcripts overlap, events will be binned into 'local'
# bug fix: order of 'cnv' in header of filtered fusion output corrected
# can report just snvs (-V) or indels (-L)
# snv.XXX output files become events.XXX
# not reporting .bed files anymore
# short form of '--lib' becomes '-l' (previously '-L')
# no more adding read support for just inversion events
# sort events by chromosome and then by coordinate before parsing genome bam; sort events by contig and then by ## coordinate before parsing contig bam
# can take a directory of bams instead of single bams
# introduce maximum of 2000 reads attempted to confirm allele and 1000 reads of positively confirmed allele - this is to get over stalls at super-deep sequenced r
# use Pysam fetch() instead of pileup() to tally reads containing allele - this is to enable above threshold to be set
# will not return depth at particular base because pileup() is not used and thresholds implemented
# fix bug when coordinate doesn't make sense in extracting reference sequences. Apply same fix in insertion detection code.
# fix bug when deletions happen on soft-clipped reads - soft-clippped regions were not accounted for in previous code
# sort events by chromosome before matching dbSNP entries
# prepare and close each chromosome for matching only once - reduced turnaround and memory