how samtools rmdup remove cut reads made more reads been drop( cut or uncut reads mapping ratio are similar)?

Question

[siqizhao2018]

compare PE cut reads with uncut reads samtools rmdup bam,how samtools rmdup remove cut reads made more reads been drop ?( cut or uncut reads mapping ratio are similar?)

[siqizhao2018]

eg:I had a PE reads,mapping then samtools rmdup,the result remove 20% reads,but when I cut the reads as 36bp(sequencing 150bp PE,36bp no any adapter),then mapping ,the mapping ratio more then 85%,but when I rmdup,it reduce 75%,only 25% reads are keep,I don't know what happend...

[finswimmer]

Hello,

the estimation of duplicates is based on mapping information. If you cut your reads to that small size of 36bp it becomes more likely that two reads map to the same position. So it is not that surprising that your duplicate number increases.

What program with which parameters are using für trimming? Why do you think this is necessary?

fin swimmer

[ATpoint]

By the way, rmdup is deprecated. For single-end reads I still use it because it is fast and can read from STDIN, but for paired-end data, you may consider markdup (introduced in SAMtools 1.6), SAMBLASTER or MarkDuplicates from Picard.