Odd stats reported for clipOverlap

[joegeorgeson]

Dear bamUtils,

I'm using bam clipOverlap --stats --poolSize 100000000 --in PE.bam --out merge.bam and get reports like the below for a handful of files. The alignments all look clean before and after, it's only the stats that don't make too much sense. Is there something I might be missing? This is for ~110bp (after cutadapt, cleaning) PE RNAseq - the imbalance between the fw/rev strands can be attributed to rRNA. The numbers I'm concerned with are the 'Average # Reference Bases Overlapped' and 'Variance of Reference Bases overlapped' as they seem abnormally high.

Overlap Statistics:
Number of overlapping pairs: 31917921
Average # Reference Bases Overlapped: 1178.49
Variance of Reference Bases overlapped: 3.93918e+07
Number of times orientation causes additional clipping: 11373
Number of times the forward strand was clipped: 22035804
Number of times the reverse strand was clipped: 9882117
WARNING: did not find expected overlapping mates for 173173 records.
Completed ClipOverlap Successfully.