Hello everyone. I'm working with STAR aligner to map microRNA seq samples into the whole mouse genome. While taking a look at the parameters, I noticed the parameter --outFilterMismatchNmax had as default mismatch number 10 (according to the STAR manual 2.7.0a). As I'm working with very small RNA, of about 20 bp length, I thought it would not make sense to allow such a high number of mismatches, so I limited it to 0 to see how the result changed. However, to my surprise, the mapping % was barely different. When using the default parameter, for my example sample I obtained an STAR output of 58.63% of uniquely mapped reads, and a 40.80% of multi-mapped reads. When limiting the mismatches to 0, the mapping was 58.00% and 41.23%, respectively. All the other parameters were left the same. How can I make sense of these results? I get almost the same mapping results, so is the maximum mismatch not that important to filter, at least for my data? Should I go on with the default STAR parameter? Am I missing something as to how this parameter works?
I would really appreciate if anyone has an idea of why this is happening or has encountered a similar issue and can help me decide about how to proceed. Thank you in advance!
I think what needs to be established how many mismatches do the initial alignments have. Take your original alignment and look for the tag called NM (NM:i:0) for each alignment. That tag ought to tell you how many mismatches do your alignments have initially. Perhaps that number was mostly 0 hence reducing the filter to 0 had no effect.