If a read had an existing BAQ tag, was clipped by our engine, and couldn't have the BAQ recalculated (for whatever reason), then we would
fail in the BQSR because we would default to using the old tag (which no longer matched the length of the read bases).
The right thing to do here is to remove the old BAQ tag when RECALCULATE and ADD_TAG are the BAQ modes used but BAQ cannot be recalculated.
Added a unit test to ensure that the tags are removed in such a case.
The check is performed by a Read Transformer that samples (currently set to once
every 1000 reads so that we don't hurt overall GATK performance) from the input
reads and checks to make sure that none of the base quals is too high (> Q60). If
we encounter such a base then we fail with a User Error.
* Can be over-ridden with --allow_potentially_misencoded_quality_scores.
* Also, the user can choose to fix his quals on the fly (presumably using PrintReads
to write out a fixed bam) with the --fix_misencoded_quality_scores argument.
Added unit tests.
-- Old code required qual to be <64, which isn't strictly necessary. Now uses the Picard SAMUtils.MAX_PHRED_SCORE constant
-- Unittest to enforce this behavior
Eric Banks
Mauricio Carneiro
Mark DePristo
David Roazen