Goby

- Fix for https://github.com/CampagneLaboratory/goby/issues/3
- Upgrade commons-io and dsiutils to latest jar versions. Log messages when scanning reads file with cfs mode.
- DistinctValueCounterBitSet: now grows to biggest size at construction time.
- Fixed a performance problem. When reading large reads file (>10GB), performance of ReadsReader would degrade
over time. This was due to caching of data in static protobuf methods of ReadCollection. We now create a
builder instance that gets garbage collected when it is no longer used. This fixes a subtle performance
problem. The same fix has been applied to alignment readers.

- concatenate-alignments mode: add ability to restrict output to a genomic slice (see -s and -e options).
- API change: AlignmentSliceHelper makes it easier to parse and process genomic slices for sets of alignments.
- concatenate-alignments mode: now transfers read groups to output in the same way that non-sorted concat does.
- concatenate-alignments mode: Add a mechanism to override/define read groups/read origin info on the fly when
reading alignments that did not include them. Coupled with changes to compact-to-sam, this makes it possible
to get BAM files with read groups directly from Goby alignments.
- compact-to-sam mode: fixed output of read groups, which were not correctly written for platform, platform unit,
and library.
- suggest-position-slices: add --restrict-per-chromosome option. When this switch is provided, slices will be
restricted to start and end on the same chromosome. This is useful to produce intervals to give Mutect,
for instance.
- Trim mode: add --trim-left --trim-right parameters to control trimming of specific sequence extremities.
- Trim mode: add --verbose flag.

- FDR mode: add ability to read groups from VCF file and adjust columns/fields marked as p-value. Mark adjusted
columns with group q-value.
- Somatic variation output format: annotate somatic p-value column with 'p-value' group. Fix the type of the p-value
column to be a number (was String in release 2.2).
- Somatic variation output format: handle unrecognized sample-ids in the parents column.
- discover-sequence-variants mode: add assertion to give hint to user that syntax is incorrect in for -s and -e options.
- compact-file-stats mode: print progress when scanning reads files. Use a buffered reader to improve read file
parsing performance.
- discover-sequence-variants: adjust multiplier for left-over filter for somatic variations output format.
- discover-sequence-variants: Add a new filter to remove indels at a site where a sample shows lots of distinct
possible indels. Indels at these sites are very likely to be artefactual. We count the number of samples where
three distinct indel genotypes are seen. If more than 1/4 of the samples have likely indel artifacts, we remove
all indel candidates at the site. maxIndelPerSite:Maximum number of distinct indels at a given genomic site.:1
Additional filter: fractionOfSamples: Maximum fraction of samples that can have an indel candidate for the indel
to be considered (indel candidates that occur in many samples are more likely to be spurious).:0.25
This filter is added to the somatic variations output format. See dynamic options for this filter with --x-help

- Remove threshold effects when calling genotypes in several samples. Modified the filters to not remove bases in
specific samples when the genotype survived filters in at least another sample (previous versions reported these
threshold edge effects as differences, which could be confusing, this version simply shows the marginal raw base
counts in samples where the genotype could have been filtered by a filter, which makes it easier to compare the
strength of the genotype support across samples). This adjustment was done for both base genotype and indel genotypes.
- LeftOverFilter: now uses minVariationSupport as minimum threshold.
- Mode suggest-position-slices: add option number-of-bytes to suggest slices with a uniform number of compressed
bytes. This option aims to provide more balanced slices in bases where the genome as very non uniform coverage
by position. With this option, the number of slices is determined to yield slices that need to decompress about
the amount of bytes indicated on the command line. `
- Framework API change: introduce class PositionToBasesMap<T> to use as type for positionToBases. The class provides
methods to get the range of positions described in the map. This unfortunately requires changes to all clients/
implementations of IterateSortedAlignments<T>.
- Mode discover-sequence-variants: Fix various problems that prevented reporting genotypes for deletions (i.e., C/-).
- Fix a potential NPE in GroupAssociations when samples are null.
- Fix for issue 2, see https://github.com/CampagneLaboratory/goby/issues/2
- Expose comparator in SortedAnnotations.

- Upgrade xstream to version 1.4.3. This fixes the compatibility problem seen when running goby 2.1.1 with java 1.7+.
Goby 2.1.2 should run with Java 1.7+, but more testing will be needed to rule out other migration problems. If you
are running JDK 1.7+ please let us know any issues you encounter.
- Fix VCFParser issue https://github.com/CampagneLaboratory/goby/issues/1. The issue could be triggered when the FORMAT
column changed from line to line.
- VCFWriter: improve support for VCF group associations. The Goby VCF parser makes it possible to associate columns
to groups (these associations are written in a FieldGroupAssociations field).
- Methylation rate VCF output: mark the context column with group 'indexed'.
- Do not try to upgrade alignments when reading the header to concatenate permutations. This is not necessary and can
open too many files when we are trying to concatenate alignments.

- Add extract-splicing-events mode. This mode is used by GobyWeb 1.9 to extract splicing events from spliced
Goby alignments (generated either by GSNAP or STAR at this time).
- Trim mode:Fix bug that caused quality scores to be duplicated (the bug triggered the assertion that checks
that sequence length equal quality length).
- Trim mode: Some sequence must remain after trimming to append to the output.
- Fix bug in alignment-to-annotation-counts when counts would be zero for samples whose name contained a
period '.' The code was incorrectly stripping alignment extensions twice.
- alignment-to-annotation-counts: add comparison description to t-test statistic column name (e.g. t-test[A/B] rather
than t-test). This change makes it possible to retrieve the t-test p-values when more than one comparison is
performed.
- Fix a bug where RandomAccessAnnotations could return results on a different chromosome.
- Add annotation loading test and fix for when annotation file is truncated. Goby now loads annotations up to
the truncation and logs truncated lines.
- Correct calculation for fold-change-magnitude column in goby diff exp mode. Previous calculation under-estimated
magnitude when comparing low rpkms.
- Fix a problem where AlignmentReaderImpl.canRead would return true when the file ended with an incorrect extension
(this problem could create subtle issues when the goby tried to access .info.txt files on a web server that did not
return 404 errors for missing content).