Entering edit mode
7.0 years ago
Tania
▴
180
Hi All
If the insert size is small, just few bases (5-8 in average) longer than a single read length, the reads overlap with 135 bases out of 150. Can this cause low mapping rate? Like if the paired reads are overlapping as if they are one single read, could this be a reason for low mapping? I am having ~70 alignment, changed the alignment algorithm, still ~70%. reads quality are ok, and no contamination when I checked. I checked adapters are removed. Could the insert size short length affect the mapping?
70% alignment may not be bad. What kind of data is this? Can you take some of the reads that don't map and blast them at NCBI to see what they are? With that much overlap you could just take one of the reads and get more or less identical information as taking both.
Human data. I will blast unmapped reads thanks. So what is the purpose of having paired reads if I can use single read and get kinda same results? Like should I argue with the company about this. I mean having paired reads is better I think, and we don't have this advantage by having too much overlap? Right?
Without knowing how the libraries were prepped and what kind of samples these are I will only comment in general. Most times libraries are made to have ~350-450 bp inserts so the PE reads will not overlap in the middle providing spatial information for the fragment. Since your insert sizes are so short you are not getting that benefit. If the provider made the libraries then you certainly could have a case to have them remake the libraries and re-sequence them, depending on the understanding/contract you have with them.