[Samtools-help] dup behaviour for samtools vs picard

[Sendu B. <sb...@sa...>]

When you do something silly like merge a bam with itself, why doesn't an 
rmdup/MarkDuplicates see the duplicate reads as duplicates? And why do 
samtools and picard behave so differently in this case?

For example, starting with a 2000 read bam file that both samtools and 
picard agree has 6 duplicate reads, merged with itself it becomes 4000 
reads. Then on the merged bam:

picard-tools MarkDuplicates marks 28 reads as duplicates.

samtools rmdup gets rid of 1073 reads.

I'd have naively expected 2006 reads to be seen as duplicates. Or 
perhaps 12. Or even 6 again. But those results just seem random?


-- 
 The Wellcome Trust Sanger Institute is operated by Genome Research 
 Limited, a charity registered in England with number 1021457 and a 
 company registered in England with number 2742969, whose registered 
 office is 215 Euston Road, London, NW1 2BE.