Still experimental. As of now, it's not useful. More analysis is needed to determine how to handle cases where UG is unsure whether a sample is het or hom.
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2442 348d0f76-0448-11de-a6fe-93d51630548a
Also updating other IntervalCleanerIntegrationTest failures that were masked by my first patch.
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2440 348d0f76-0448-11de-a6fe-93d51630548a
1) sam-jdk apparently no longer supports custom tags with type int[] values.
2) BAM output for indel cleaner integration test changed in a way that's so subtle it can't be seen after converting the output to .sam.
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2439 348d0f76-0448-11de-a6fe-93d51630548a
1: all overlapping and abutting intervals merged (ALL),
2: just overlapping, not abutting intervals (OVERLAPPING_ONLY),
3: no merging (NONE). This option is not currently allowed, it will throw an exception. Once we're more certain that unmerged lists are going to work in all cases in the GATK, we'll enable that.
The command line option is --interval_merging or -im
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2437 348d0f76-0448-11de-a6fe-93d51630548a
1. allele balance annotation is now weighted by genotype quality (so we don't get misled by borderline het calls)
2. Updates to the Unified Genotyper for parallelization:
a. verbose writing now works again; arg was moved from UAC to UG
b. UG checks for command that don't work with parallelization
c. some cleanup
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2432 348d0f76-0448-11de-a6fe-93d51630548a
The cycle covariate is now first/second of pair aware. I'm taking it on faith from both Chris Hartl (waiting on slides from him) and Tim that this is the right thing to do. We'll have Ryan confirm it all next week.
The only change is that if a read is the second of a pair, we multiple the cycle by -1 (a simple way of separating its index from that of its mate).
Of course, this broke all integration tests.
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2431 348d0f76-0448-11de-a6fe-93d51630548a
depending on its output format. Current implementation is probably a bit overkill-ish and
we can whittle this down to what's absolutely necessary.
Writing VCFs to the 'out' protected printstream may not work at this moment.
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2425 348d0f76-0448-11de-a6fe-93d51630548a
Also, slightly optimized the cleaner by using readBases (instead of readString) and caching cigar element lengths.
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2419 348d0f76-0448-11de-a6fe-93d51630548a
this call suggests that I may be thinking about the typing of the GenotypeWriter object the wrong way.
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2418 348d0f76-0448-11de-a6fe-93d51630548a
Added - A paired read order covariate to use with recalibration. Currently experimental: for instance, what's a proper pair versus just a pair? Nobody should use this one...
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2401 348d0f76-0448-11de-a6fe-93d51630548a
Adding first pass of stub and storage classes for the GenotypeWriters so that UG can be parallelizable. Not hooked up yet, so UG is unchanged.
The mergeInto() code in the storage class is ugly, but it's all Tribble's fault. We can clean it up later if this whole thing works.
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2400 348d0f76-0448-11de-a6fe-93d51630548a
2. isNoCall() added to Genotype interface so that we can distinguish between ref and no calls (all we had before was isVariant())
3. Added Hardy-Weinberg annotation; still experimental - not working yet so don't use it.
4. Move 'output type' argument out of the UnifiedArgumentCollection and into the UnifiedGenotyper, in preparation for parallelization.
5. Improved some of the UG integration tests.
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2398 348d0f76-0448-11de-a6fe-93d51630548a
shard for cases when the index file isn't available. Works for the case in question, but is not
guaranteed to work in general. Will be replaced once the new sharding system comes online.
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2383 348d0f76-0448-11de-a6fe-93d51630548a
1. push the ReadBackedPileup filtering up into the ReadFilters for read-based filters
2. stop querying the cigar for its length (just do it once)
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2381 348d0f76-0448-11de-a6fe-93d51630548a
Reworked all of the integration tests so that they're now more comprehensive, cover more of what we wan to test, and don't take forever to run.
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2376 348d0f76-0448-11de-a6fe-93d51630548a
Bad/suspicious bases/reads (high mismatch rate, low MQ, low BQ, bad mates) are now filtered out by default (and not used for the annotations either), although this can all be turned off.
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2373 348d0f76-0448-11de-a6fe-93d51630548a
Changed peek() to element() to be consistant with the Java standards of the Queue and Stack classes (element() throws an exception if a record isn't available).
Also updated some of the ROD iterator next() methods to throw NoSuchElementException if next() is called when a record isn't available.
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2372 348d0f76-0448-11de-a6fe-93d51630548a
This is a temporary fix - pushed more elegant solution over to Matt.
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2360 348d0f76-0448-11de-a6fe-93d51630548a
[Also, enable Mark's new UG arguments]
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2355 348d0f76-0448-11de-a6fe-93d51630548a
1. Initial code for annotating calls with the base mismatch rate within a reference window (still needs analysis).
2. Move error checking code from rodVCF to VCFRecord.
3. More improvements to SNP Genotype callset concordance.
4. Fixed some comments in Variation/Genotype
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2341 348d0f76-0448-11de-a6fe-93d51630548a
below the specific command-line arguments for the walker. Also introduced
@help.summary to override summary descriptions if required.
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2337 348d0f76-0448-11de-a6fe-93d51630548a
Also increased the line width as much as I could tolerate (100 characters -> 120 characters).
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2336 348d0f76-0448-11de-a6fe-93d51630548a
The only major difference is that we are now able to get accurate allele balance ratios.
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2321 348d0f76-0448-11de-a6fe-93d51630548a
This is basically a wrapper class around AlignmentContext which allows you to stratify a context by e.g. reads on forward vs. reverse strands.
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2312 348d0f76-0448-11de-a6fe-93d51630548a
Convert it from a completely wonky solution to a slightly less wonky solution
that will work in more cases.
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2310 348d0f76-0448-11de-a6fe-93d51630548a
- Hook up Variation and Genotype in SSG
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2300 348d0f76-0448-11de-a6fe-93d51630548a
-Make minConfidenceScore in VariantEval a double so non-integer values can be used (requested by Steve H).
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2290 348d0f76-0448-11de-a6fe-93d51630548a
- Removed MQ0 annotation
- Updated RMS MQ annotation to use new pileup
- UG now outputs all of its arguments as key/value pairs in the header (for VCF)
- Cleaned up VCFGenotypeWriterAdapter interface a bit
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2288 348d0f76-0448-11de-a6fe-93d51630548a
We are now VCF3.3 compliant.
(Only a few more stages left. Sigh.)
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2287 348d0f76-0448-11de-a6fe-93d51630548a
feedback received. Also, add a flag to build.xml to disable generation of
docs on demand (use ant -Ddisable.doc=true to disable docs).
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2284 348d0f76-0448-11de-a6fe-93d51630548a
VCFRecord now implements Variation, VCFGenotypeRecord now implements Genotype.
Because of this change, RodVCF is now just a wrapper around the VCFRecord and does nothing else. Also, one can call toVariation on the VCFGenotypeRecord and it returns the VCFRecord.
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2271 348d0f76-0448-11de-a6fe-93d51630548a
This stage consists only of the code originating in the Genotyper and flowing through to the genotype writers. I haven't finished refactoring the writers and haven't even touched the readers at all.
The major changes here are that
1. Variations which are BackedByGenotypes are now correctly associated with those Genotypes
2. Genotypes which have an associated Variation can actually be associated with it (and then return it when toVariation() is called).
The only integration tests which need to be updated are MSG-related (because the refactoring now made it easy for me to prevent MSG from emitting tri-allelic sites).
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2269 348d0f76-0448-11de-a6fe-93d51630548a
and displays walker data extracted from the JavaDoc. Needs a bit of help,
both in content and flexibility of package naming.
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2267 348d0f76-0448-11de-a6fe-93d51630548a
in GenomeAnalysisTK.jar. Still no support for actually displaying the archived javadoc. Also change the approach
to providing package javadocs: retired the deprecated package.html file in favor of Java1.5-style package-info.java.
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2263 348d0f76-0448-11de-a6fe-93d51630548a
-added code to cache the allele list so it didn't need to get recomputed each time it was requested.
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2260 348d0f76-0448-11de-a6fe-93d51630548a
2. Fixed bug in histogram calculation for small intervals
3. Better output in DoCWalker
4. Comments added to code
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2245 348d0f76-0448-11de-a6fe-93d51630548a
I am now at a point where I can merge the functionality from other coverage walkers into this one.
Thanks to Andrew for input.
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2239 348d0f76-0448-11de-a6fe-93d51630548a
- ClipReads
- PrintRODs (generalized to print all RODs that are Variations)
- FixBAMSortOrderTag (added documentation to walker so that people know what it does and why)
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2238 348d0f76-0448-11de-a6fe-93d51630548a
- Moved validation walkers to new qc dir
- Killed unused test
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2218 348d0f76-0448-11de-a6fe-93d51630548a
- Use QualityUtils when phred-scaling now
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2217 348d0f76-0448-11de-a6fe-93d51630548a
- Renamed keys to fit with the standard naming
- FisherStrand is no longer standard
- Integration tests no longer test experimental annotations since they're not stable
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2216 348d0f76-0448-11de-a6fe-93d51630548a
-Renamed methods to be more consistent.
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2214 348d0f76-0448-11de-a6fe-93d51630548a
- Added implementation of -bySample in DoCWalker
- Removed CoverageBySample and added a watered down version to the examples directory
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2209 348d0f76-0448-11de-a6fe-93d51630548a
Moved/killed a few other walkers (with permission).
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2204 348d0f76-0448-11de-a6fe-93d51630548a
- DepthOfCoverage is now by reference (so locus-by-locus output correctly reports zero-coverage bases)
- VariantsToVCF now lets you bind variants with any string except intervals and dbsnp (not just NA######)
- A PileupWalker integration test on a particularly nasty FHS site
- Two second-base annotation related integration tests on that same site
+ outputs were all hand-validated in matlab; within a certain tolerance for the annotations
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2197 348d0f76-0448-11de-a6fe-93d51630548a
Finished converting genotyper and annotator code to new ReadBackedPileup system.
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2192 348d0f76-0448-11de-a6fe-93d51630548a
I note that all integration/unit tests pass except for BaseTransitionTableCalculatorJava, which is already broken.
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2182 348d0f76-0448-11de-a6fe-93d51630548a
Also, Variations should be INSERTIONs or DELETIONs (and not just INDELs).
Technically, VCF records can be indels now.
More changes coming
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2150 348d0f76-0448-11de-a6fe-93d51630548a
1. Only do calculations in UG for alternate allele with highest sum of quality scores (note that this also constitutes a bug fix for a precision problem we were having).
2. Avoid using Strings in DiploidGenotype when we can (it was taking 1.5% of my compute according to JProfiler)
UG now runs in half the time for JOINT_ESTIMATE model.
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2141 348d0f76-0448-11de-a6fe-93d51630548a
[As a courtesy I fixed all instances once I was updating GenotypeLikelihoods]
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2136 348d0f76-0448-11de-a6fe-93d51630548a
Computes a metric for how much error is left that isn't explained by ref or snp bases. This is the sum of Q scores, weighted by the proportion of non-ref non-snp bases to non-snp bases. Reported in Log space.
Update to the integration test so bamboo doesn't look as though someone murdered it with a spork
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2124 348d0f76-0448-11de-a6fe-93d51630548a
Added integration test for pooled model and updated other joint estimation tests to be more comprehensive now that they are faster.
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2123 348d0f76-0448-11de-a6fe-93d51630548a
2. UG: don't print slod if lods are infinite (todo: figure out a good guess instead)
3. UG: if probF=0 for 2 alt alleles are both 0 (because of precision), use log values to discriminate
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2116 348d0f76-0448-11de-a6fe-93d51630548a
2. Retired allele frequency range from UG, which wasn't very useful.
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2113 348d0f76-0448-11de-a6fe-93d51630548a
- For all reference bases, the proportion of their second best bases that support the SNP
- the proportion of non-reference bases that support the SNP
and reports the difference between the two. Initially I was taking depth into account as well, but that did not appear to work as nicely as I'd like (even at 20,000x depth, if 95% of the non-reference bases are C, and 98% of the reference second-best-bases are C, then we would want to be suspicious of it; but perhaps slightly less so than if the depth were only 20...)
Anyway it's now available. I'm not sure how useful it will be, but I spawned the FHS annotation jobs again, so we'll see.
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2109 348d0f76-0448-11de-a6fe-93d51630548a
[There was no reason to enforce that every VCF being output from the GATK should have the samples sorted, since someone might want them ordered non-alphabetically]
git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2102 348d0f76-0448-11de-a6fe-93d51630548a
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
ebanks
hanna
aaron
alecw
depristo
rpoplin
chartl
asivache
andrewk
kiran