[Bioc-devel] ShortRead: 'qa' fails for single read alignments

Hi,

while using the 'ShortRead' package for some quality assessment of
aligned reads (see example below), I observed the following behavior:

## Example code ##

library(ShortRead)
qa1 <- qa(dirPath="tmp/", pattern="*sub.bam", type="BAM")
report_html(qa1, dest="out")

##

1. For R-2.14.0, the report is built as expected (see
http://www.ebi.ac.uk/~jgehring/share/shortRead-pkg/perCycleBaseCall-R-2.14.0.pdf
for a comparison).

2. For R-2.15.0, the cycle-specific base calls and read quality plot
looks mixed up (see
http://www.ebi.ac.uk/~jgehring/share/shortRead-pkg/perCycleBaseCall-R-2.15.0.pdf).


3. For R-2.16.0devel (2012-05-24 r59439), the 'qa' command fails with
the error message:
""
Error: ValueUnavailable
0 elements returned; expected >=1
In addition: Warning message:
UnspecifiedWarning
elements: 1 2 3 4
UnspecifiedError: bamFlagTest(flag, "isValidVendorRead")
'is' must be character(1) in 'isPaired' 'isProperPair' 'isUnmappedQuery'
'hasUnmappedMate' 'isMinusStrand' 'isMateMinusStrand' 'isFirstMateRead'
'isSecondMateRead' 'isNotPrimaryRead' 'isNotPassingQualityControls'
'isDuplicate'
UnspecifiedError: bamFlagTest(flag, "isValidVendorRead")
'is' must be character(1) in 'isPaired' 'isProperPair' 'isUnmappedQuery'
'hasUnmappedMate' 'isMinusStrand' 'isMateMinusStrand' 'isFirstMateRead'
'isSecondMateRead' 'isNotPrimaryRead' 'isNotPassingQualityControls'
'isDuplicate'
UnspecifiedError: bamFlagTest(flag, "isValidVendorRead")
'is' must be character(1) in 'isPaired' 'isProperPair' 'isUnmappedQuery'
'hasUnmappedMate' 'isMinusStrand' 'isMateMinusStrand' 'isFirstMateRead'
'isSecondMateRead' 'isNotPrimaryRead' 'isNotPassingQualityControls'
'isDuplicate'
UnspecifiedError: bamFlagTest(flag, "isValidVendorRead")
'is' must be character(1) in 'isPaired' [... truncated]
""

See also
- http://www.ebi.ac.uk/~jgehring/share/shortRead-pkg/session-info-2.14.txt
- http://www.ebi.ac.uk/~jgehring/share/shortRead-pkg/session-info-2.15.txt
- http://www.ebi.ac.uk/~jgehring/share/shortRead-pkg/session-info-2.16.txt
for the corresponding session infos.

Can this be caused by having BAM files with single-read alignments?
Also, I'm not sure if the different behavior for R-2.15 and R-2.16 is
directly related.


Best
Julian