Updated the integration tests that were failing to due to different ordering of genotyping entries in VCF, I'll check in the VCF diff tool I wrote when I get a cycle or two.
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2092 348d0f76-0448-11de-a6fe-93d51630548a
-If qscore is infinity (because of precision) make a best guess instead
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2076 348d0f76-0448-11de-a6fe-93d51630548a
2. Allele frequency spectrum is not emitted for single samples (since it doesn't make sense).
3. If in pooled mode, throw an exception of pool size isn't set appropriately.
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2072 348d0f76-0448-11de-a6fe-93d51630548a
We totally *do* want to annotate the call if called by another walker. Totally boneheaded misenterpretation of what the code was doing -- Eric, please forgive me for being an idiot.
Instead, change the StingException to what it really should be -- an IllegalStateException, which is not coincidentally already handled by the calling function.
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2067 348d0f76-0448-11de-a6fe-93d51630548a
do **not** attempt to annotate if UnifiedGenotyper is called from another walker! Why this didn't break the build earlier I have no idea.
Ultimately, there should be a better way of interfacing UG with another walker -- what if some other walker wants the annotations from UG? But since we're calling map directly -- and the annotations don't get returned directly from map -- this needs to be handled differently, while the map function should ultimately return the LOD score or quality under the GCM alone.
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2066 348d0f76-0448-11de-a6fe-93d51630548a
Second base skew annotations and integration tests. Nothing need be given except -A SecondBaseSkew; the statistic it annotates calls with is a chi-square statistic given by the deviation of the observed proportion of reference second-best-bases from the expected 1/3. Future additions may be to ask that the deviation be instead from a given transition table.
A big note for all users: All IllegalStateExceptions from the variation ROD (e.g. the RodGeliText) are dealt with SILENTLY. I understand this isn't optimal, but I'd rather simply not annotate a non-bi-allelic site than fail completely (there are quite a few such sites even on the regions over which the integration test has been written).
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2064 348d0f76-0448-11de-a6fe-93d51630548a
-Use phred-scale for fisher strand test
-Use only 2N allele frequency estimation points
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2059 348d0f76-0448-11de-a6fe-93d51630548a
VariantAnnotator can be called as a standalone walker or by another walker, as it is by the UnifiedGenotyper. UG now no longer computes any of this meta data - it relegates the task completely to the annotator (assuming the output format accepts it).
This is a fairly all-encompassing check in. It involves changes to all of the UG code, bug fixes to much of the VCF code as things popped up, and other changes throughout. All integration tests pass and I've tediously confirmed that the annotation values are correct, but this framework could use some more rigorous testing.
Stage 2 of the process will happen later this week.
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2053 348d0f76-0448-11de-a6fe-93d51630548a
2. Fix for Kiran: allow UG to call SNPs at deletion sites; we'll add an annotation to the VariantAnotator for deletions at the locus (next week).
3. Added integration tests for joint estimation model
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2038 348d0f76-0448-11de-a6fe-93d51630548a
2. Don't print verbose output from SLOD calculation (it's just a repeat of previous output).
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2032 348d0f76-0448-11de-a6fe-93d51630548a
Just stomped on the existing md5s because that's what Eric told me to do.
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2029 348d0f76-0448-11de-a6fe-93d51630548a
1. Add dbsnp RS ID to VCF output from genotyper; to do this I needed to fix the dbsnp rod which did not correctly return this value.
2. Remove AlleleBalanceBacked and instead generalize the arbitrary info fields backing VCFs (and potentially others) in preparation for refactoring VariantFiltration next week.
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2028 348d0f76-0448-11de-a6fe-93d51630548a
Revert until we find out whether the cause is legit.
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2017 348d0f76-0448-11de-a6fe-93d51630548a
1. Don't cap q-scores at 99
2. Scale SLOD to allow more resolution in the output
3. UG outputs weighted allele balance (AB) and on-off genotype (OO) info fields for het genotype calls (works for joint estimation model and SSG)
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2011 348d0f76-0448-11de-a6fe-93d51630548a
-Don't restrict info fields to 2-letter keys
[about to move these to core]
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2002 348d0f76-0448-11de-a6fe-93d51630548a
-Improvement to snp genotype concordance test
And with that, it looks like I get revision #2000.
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2000 348d0f76-0448-11de-a6fe-93d51630548a
alignment start will confuse the sharding system and cause it to return duplicate reads.
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@1987 348d0f76-0448-11de-a6fe-93d51630548a
It needs to be modified a bit and then hooked up to a pooled model, but that is now possible.
At this point, there is no difference to the Unified Genotyper.
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@1978 348d0f76-0448-11de-a6fe-93d51630548a
Added a PlusOneFixIterator that wraps other iterators, and eliminates reads that start outside of our intended interval (interval stop - 1). Updated and checked BamToFastqIntegrationTest MD5 sums.
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@1976 348d0f76-0448-11de-a6fe-93d51630548a
Now, all output is generalized and all of the intelligence lies where it is supposed to.
Next stage is syncing up old and new models and making sure we're outputting exactly what we should.
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@1960 348d0f76-0448-11de-a6fe-93d51630548a
-Make rods return the appropriate type of Genotype calls from getGenotype().
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@1954 348d0f76-0448-11de-a6fe-93d51630548a
-Do the right thing in all models for all-base-mode (for Kiran).
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@1940 348d0f76-0448-11de-a6fe-93d51630548a
Some things still need to be changed, but it will entail some more design decisions first (which means I get to bug M&A again tomorrow!).
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@1930 348d0f76-0448-11de-a6fe-93d51630548a
relationship between these two classes needs to be rethought; see JIRA
GSA-207.
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@1904 348d0f76-0448-11de-a6fe-93d51630548a
The whole GenotypeCall framework needs to be changed, but this will work for the time being.
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@1902 348d0f76-0448-11de-a6fe-93d51630548a
-Don't print verbose/debugging output to logger, but instead specify a file in the argument collection (and then we only need to print conditionally)
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@1899 348d0f76-0448-11de-a6fe-93d51630548a
-Allow walkers calling the UG to pass in their own argument collections
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@1896 348d0f76-0448-11de-a6fe-93d51630548a
Because it doesn't actually use EM, it's no longer a subclass of the EM model.
Note that you can't use it just yet because it doesn't actually emit calls (just prints to logger). I need to deal with general UG output tomorrow. Hold off until then, Mark, and then you can go wild.
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@1891 348d0f76-0448-11de-a6fe-93d51630548a
Chris, if this breaks an integration test, you get it.
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@1879 348d0f76-0448-11de-a6fe-93d51630548a
@PoolUtils - split reads by indel/simple base
@BaseTransitionTable - complete refactoring, nicer now
@UnifiedArgumentCollection - added PoolSize as an argument
@UnifiedGenotyper - checks to ensure pooled sequencing uses the appropriate model
@GenotypeCalculationModel - instantiates with the new PoolSize argument
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@1867 348d0f76-0448-11de-a6fe-93d51630548a
Also moved a buch of Lists over to Sets for consistancy.
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@1859 348d0f76-0448-11de-a6fe-93d51630548a
also: their, I hope your happy Eric, from now on I'll try not to flout my awesomest grammer in the future accept when I need to illicit a strong response :-)
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@1858 348d0f76-0448-11de-a6fe-93d51630548a
We need to filter contexts in that case since the calling walkers don't get UG's traversal-level filters.
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@1848 348d0f76-0448-11de-a6fe-93d51630548a
-Add another filter for read groups for Chris
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@1835 348d0f76-0448-11de-a6fe-93d51630548a
Remove all sequenom stuff from the FastaAlternateReferenceMaker so it can just concentrate on making alternate references...
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@1831 348d0f76-0448-11de-a6fe-93d51630548a
We currently get identical lods and slods as MultiSampleCaller (except slods for ref calls, as I discussed with Jared) and are a bit faster in my few test cases. Single-sample mode still emulates SSG.
The remaining to do items:
1. more testing still needed
2. we currently only output lods/slods, but I need to emit actual calls
3. stubs are in place for Mark's proposed version of the EM calculation and now I need to add the actual code.
More check-ins coming soon...
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@1821 348d0f76-0448-11de-a6fe-93d51630548a
For now, a single sample input will be special-cased in the EM model - but that will change when the EM model degenerates to the single sample output with a single sample as input. For now, the EM code for multi-samples isn't finished; I'm planning on checking that in soon.
The SingleSampleIntegrationTest now uses the UnifiedCaller instead of SSG, and so should all of you. More on that in a separate email.
Other minor cleanups added too.
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@1785 348d0f76-0448-11de-a6fe-93d51630548a
Also, with Picard's latest changes, we need to make sure we don't double-close the sam writer.
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@1779 348d0f76-0448-11de-a6fe-93d51630548a
-Adding initial version of Multi-sample calculation model. This still needs much work: it needs to be cleaned up and finished. Right now, it (purposely) throws a RuntimeException after completing the EM loop.
Also:
-Fix logic in GenotypeLikelihoods.setPriors
-Add logger to the models for output
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@1764 348d0f76-0448-11de-a6fe-93d51630548a
Also a fix to the GLF tests, and a correction to PrintReadsWalker to remove the close() on the output source, the source handles that itself (and you get a double close).
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@1758 348d0f76-0448-11de-a6fe-93d51630548a
Some notes:
1. This design should be flexible enough to include pooled calling (for now) after discussions with Chris.
2. Using the unified caller with the SingleSampleCalculationModel emits the exact same output as SSG over all of chr20 for NA12878. Additionally, when we include the "max deletions allowed at a locus" argument (so we don't try to call SNPs at deletion sites), it removed 233 SNP calls in chr20 that were clearly indel artficts.
3. The MultiSampleEMCalculationModel is still a work in progress and will be checked in later this week.
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@1740 348d0f76-0448-11de-a6fe-93d51630548a
rpoplin
aaron
alecw
ebanks
hanna
depristo
chartl
asivache
kiran
andrewk