after piccard MarkDuplicates drop in mapping percentage

[trotos]

java -Xmx14g -jar MarkDuplicates.jar INPUT=file.sorted.bam OUTPUT=file.noDUP.bam REMOVE_DUPLICATES=TRUE METRICS_FILE=metrics2.log
I'me running the above command.
Flagstat on the sorted bam is, before removal of dublicates

34487707 + 0 in total (QC-passed reads + QC-failed reads)
0 + 0 duplicates
31603946 + 0 mapped (91.64%:-nan%)
0 + 0 paired in sequencing
0 + 0 read1
0 + 0 read2
0 + 0 properly paired (-nan%:-nan%)
0 + 0 with itself and mate mapped
0 + 0 singletons (-nan%:-nan%)
0 + 0 with mate mapped to a different chr
0 + 0 with mate mapped to a different chr (mapQ>=5)

after removal:

9371101 + 0 in total (QC-passed reads + QC-failed reads)
0 + 0 duplicates
6487340 + 0 mapped (69.23%:-nan%)
0 + 0 paired in sequencing
0 + 0 read1
0 + 0 read2
0 + 0 properly paired (-nan%:-nan%)
0 + 0 with itself and mate mapped
0 + 0 singletons (-nan%:-nan%)
0 + 0 with mate mapped to a different chr
0 + 0 with mate mapped to a different chr (mapQ>=5)

What is going on????

usually there is no significant reduction of mappability (0,1-0,3)


Thank you in advance

[Brian Bushnell]

These numbers are a bit extreme, but there's nothing surprising. In each case, there were 2883761 reads that would not map, so the "mappability" did not change. Since duplicates are highly likely to be error-free and originate from the dominant organism, duplicate removal will selectively remove the reads that are most likely to map.

You probably have 2883761 reads with high error rates, or adapter sequence, or from various contaminant organisms. I suggest BLASTing some of them to see what they came from, or aligning with a more sensitive aligner if you want to increase the mapping rate.