simulated 50 bp SE gives higher alignment rate compared to 100 bp PE

[alyamahmoud]

Hi All

I have 100 bp PE ss RNASeq reads. I tried to simulate 50 bp SE out of those by trimming using trimmomatic and then mapping both 100 PE and simulated 50 bp SE using tophat. I used -g 1 to make sure I get reads counted only once.

Comparing the alignment rate; the 50 bp SE have higher number of mapped reads in the accepted_hits.bam output of tophat. I also checked the correlation between log2 fold change of DEG and they seem to be very well correlated between the two experiments.

Can anyone explain to me why would 50 bp SE reads map better than 100 bp PE ?

Thanks a lot
Alyaa

[Dario1984]

It's because for single-end reads you only have 50 bases to map, whereas for paired-end you have to map two sets of 50 bases correctly. Assuming sequencing errors happen at the same rate, there would be twice as many errors to overcome for mapping the paired-end data.