Entering edit mode
5.3 years ago
devikaparvathy ▴ 50
I am using featurecounts after mapping of my reads. I have a paired-end sequencing data but Only a low percentage of reads were mapped when the paired-end reads were mapped separately and hence they were merged to a single-end read and mapped.
I know that when we use paired-end reads in featurecounts, we have to give paired-end option and also allow multi-overlap.
Does this stand true for merged single end reads too? Or should i give it as single-end read (since I have only a single bam file) and allow multi-overlaps?
The read counts are similar to ht-seq count when the multi-overlap option is set false.
multi-overlap == multi-mappedin this case?
You should not be counting reads that multi-map. It sounds like you have short inserts which are allowing your paired-end reads to largely overlap.
Why did you try to align the PE reads separately?
No, multimapping and multioverlapping are different. I checked this by running alternate featurecount analysis on the same dataset by changing each of the option to true and false and analysing the results. Multioverlapping when kept to false, produces many ambiguous, and the counts for lot of genes are given as zero (Same as HT-seq count). However, multimapping when kept to false produces a different result.
If multi-overlap is allowed (advantage of featurecounts over ht-seq count) then I get counts for some important genes (whose counts are zero otherwise) and these genes are actually known to play a role in the phenomenon that I am studying.
NOTE: I am working on a bacterial genome, so should not be multi-map reads counted?
I just tried and checked out initially the mapping percentage with the pared-end reads. Tried with different aligners but still the mapping percentage was low. then proceeded to merge them.
Is there anyone who can answer this question? I think, It is really good point for the downstream analysis. I am also in the decision step for "multioverlapping" flag in featurecounts.