Guillermo del Angel 76a95fdedf Full implementation of multiallelic exact model for pools. Still super-linear so not useable at scale but it should be a gold standard to compare to. Unit tests are not exhaustive yet, will be expanded to provide better test coverage. Small inconsequential optimization in MathUtils: we're already caching log10(factorial(n)) for large n, so might as well use the cached values to compute binomial and multinomial coefficients instead of the log-gamma approximation which is more expensive (doesn't seem to save much time either in PoolCaller nor in UG though).		2012-05-02 09:24:28 -04:00
..
R	Add warning message about ratios in variantQCreport, give ratio for MAF > 10%	2012-04-25 12:22:32 -04:00
c	At chartl's request, add the bwa aln -N and bwa aln -m parameters to the bindings.	2012-01-17 14:47:53 -05:00
chainFiles	Reorganized the codebase beneath top-level public and private directories,	2011-06-28 06:55:19 -04:00
doc	Reorganized the codebase beneath top-level public and private directories,	2011-06-28 06:55:19 -04:00
java	Full implementation of multiallelic exact model for pools. Still super-linear so not useable at scale but it should be a gold standard to compare to. Unit tests are not exhaustive yet, will be expanded to provide better test coverage. Small inconsequential optimization in MathUtils: we're already caching log10(factorial(n)) for large n, so might as well use the cached values to compute binomial and multinomial coefficients instead of the log-gamma approximation which is more expensive (doesn't seem to save much time either in PoolCaller nor in UG though).	2012-05-02 09:24:28 -04:00
keys	Public-key authorization scheme to restrict use of NO_ET	2012-03-06 00:09:43 -05:00
packages	Remove unnecessary and obscure usage of old R	2012-05-01 13:39:09 -04:00
perl	Update to the bindings for liftOverVCF.pl (to -V from -B)	2011-09-15 15:33:09 -04:00
scala	Refactoring/fixing up UG HMM code: a) Make code use PairHMM class instead of having duplicated code. That way UG and HaplotypeCaller now use same core code. Changes to be able to do this: 1. Compute context-dependent GOP as a function of read, not of haplotype, b) Extracted code to initialize HMM arrays into separate method, c) Move PairHMM class and unit test to public, d) Reenable banded code in PairHMM, inverted sense of flag (true=enable feature) but leave off in HaplotypeCaller.	2012-04-17 14:22:48 -04:00
testdata	continuing the BQSR triage...	2012-04-05 14:34:15 -04:00