Now only filtered reads are unstranded. All consensus reads have strand, so that we
emit 2 consensus reads in general now: one for each strand.
This involved some refactoring of the sliding window which cleaned it up a lot.
Also included is a bug fix:
insertions downstream of a variant region weren't triggering a stop to the compression.
1. Fix for the -okayToMiss argument for indels.
In cases where we make calls with different alleles, it wasn't allowing us to skip the site for FNs.
2. Need to add confidence and isComplexEvent attributes to the equality and duplicate checks in MVC.
3. Treat unknown confidences as reviewed for now.
We need this until IGV gets updated to use confidences for reviews.
Try to reduce the number of tests failing with file not found
errors due to random automount failures by cd'ing into a
preset list of directories at the start of each job in an
effort to trigger automount.
ant -p outputs only targets that have description attributes.
Modify build.xml so only important targets that users might actually
want to use are output by ant -p.
-User must provide a mapping file via new --sample_rename_mapping_file argument.
Mapping file must contain a mapping from absolute bam file path to new sample name
(format is described in the docs for the argument).
-Requires that each bam file listed in the mapping file contain only one sample
in their headers (they may contain multiple read groups for that sample, however).
The engine enforces this, and throws a UserException if on-the-fly renaming is
requested for a multi-sample bam.
-Not all bam files for a traversal need to be listed in the mapping file.
-On-the-fly renaming is done as the VERY first step after creating the SAMFileReaders
in SAMDataSource (before the headers are even merged), to prevent possible consistency
issues.
-Renaming is done ONCE at traversal start for each SAMReaders resource creation in the
SAMResourcePool; this effectively means once per -nt thread
-Comprehensive unit/integration tests
Known issues: -if you specify the absolute path to a bam in the mapping file, and then
provide a path to that same bam to -I using SYMLINKS, the renaming won't
work. The absolute paths will look different to the engine due to the
symlink being present in one path and not in the other path.
GSA-974 #resolve
Merged bug fix from Stable into Unstable
Conflicts:
protected/java/test/org/broadinstitute/sting/gatk/walkers/compression/reducereads/SlidingWindowUnitTest.java
Previous fixes and tests only covered trailing soft-clips. Now that up front
hard-clipping is working properly though, we were failing on those in the tool.
Added a patch for this as well as a separate test independent of the soft-clips
to make sure that it's working properly.
So, compromise solution is to go back to having biallelic PLs but emit a new FORMAT field, called APL, which has the 10 values, but all other statistics and regular PLs are computed as before.
Note that integration test had to be disabled, as the BCF2 codec apparently doesn't support writing into genotype fields other than PL,DP,AD,GQ,FT and GT.
1. All NA12878DBWalkers that export/emit sites need to do so in order; also one should be able
to use -L with them and not have it iterate over all possible sites.
Updated ExportReviews and ExtractConsensusSites to adhere to these constraints.
2. Added the option to AssessNA12878 to have it ignore FNs that overlap with a provided VCF.
This is useful if you have a list of sites from reviews that are okay to be missed in
particular techs only (because for some reason there is coverage but no evidence of the
alternate allele in them) - intended to be used with Jenkins.
3. Hooked up the logic of complex events all the way through the KB.
Now the consensus incorporates whether a call is complex and the assessor does not penalize for them.
4. Fixed long-standing bug that I managed to find accidentally:
AssessNA12878 was closing its DB connection before its final call to includeMissingCalls().
5. Hooked up the per-call confidences through the KB.
We no longer have a 2-tiered priority system in the KB (reviews and everything else) but instead
use a quasi-Bayesian estimator (will update to proper Bayesian treatment if needed).
Now ImportCallset and ImportReviews assigns confidences as appropriate.
Also needed to fix up the consensus logic for calls with UNKNOWN status.
-Two SAMReaderIDs that pointed at the same underlying bam file through
a relative vs. an absolute path were not being treated as equal, and
had different hash codes. This was causing problems in the engine, since
SAMReaderIDs are often used as the keys of HashMaps.
-Fix: explicitly use the absolute path to the encapsulated bam file in
hashCode() and equals()
-Added tests to ensure this doesn't break again
Yossi Farjoun
droazen
Eric Banks
Joseph Rose
Mauricio Carneiro
chartl
Geraldine Van der Auwera
Ryan Poplin
Louis Bergelson
sathibault
Guillermo del Angel