Add option ALIGNER_PROPER_PAIR_FLAGS to MergeBamAlignment to tell it to keep the aligner's notion of proper pair flag rather than overwriting it with MergeBamAlignment's notion.

Set SortingCollection.destructiveIteration true by default, in order to enable earlier GC of no-longer-needed SAMRecords.

Significant code refactoring in IlluminaBasecallsToSam; overhaulded threading model to improve CPU utilization; added test for Tile comparator

Augmented plots for GC Bias Metrics, Mean Quality by Cycle, Quality Score Distribution, and RNA Sequence Coverage.  Specifically, if the plot was generated from a read group .bam, the title of the plot includes the read group's library.

added method to make a read unmapped and a method to check whether any cigar element consumes both read and reference bases to SAMUtils; used these methods in AbstractAlignment merger to catch reads with cigar strings that don't consume any read or reference bases (e.g. all soft-clipped) and make them unmapped

Handle case where an alignment record has no cigar elements that consume both the read and the reference (e.g. the read is all soft-clipped)

No commit message

Add support for targetedPcrMetrics and collected common HsMetrics and TargetedPcrMetrics behavior into TargetMetricsCollector

Added a new "ProgressLogger" class that facilitates more useful and standard progress logging for any program that iterates through a stream of SAMRecords.  Adapted most command line programs to use it.

Abstract metric accumulation functionality into MultiLevelCollector and refactor all classes that use MultilevelMetrics

Switch SortingLongCollection from memory-mapped I/O to DataInput/OutputStream, in order to avoid problems due to JVM's delayed release of mapped memory.

Performance fix to EstimateLibraryComplexity.  Previously it would attempt to find duplicates within all groups of reads with a common prefix which would bog down if a technical sequence like adapter dimer dominated the library.  The fix is just to ignore groups that have > some large multiple of the expected average number of read pairs per group.  The default is 500 times over the mean.

Add option PRIMARY_ALIGNMENT_STRATEGY to MergeBamAlignment, which enables selection of the strategy for selecting the primary alignment for a read if there are multiple alignments marked as primary, or if none of the alignments for a read are marked as primary.
In addition to the existing strategy, BestMapq, add a new strategy EarliestFragment which only works for fragments, and selects the alignment that maps the earliest base in a read.