Understanding --Genome-Read-Mismatches And --Read-Mismatches Parameter In Tophat

Question

[Arun]

Dear all,

I am using currently Tophat v2.0.4 and bowtie 2.0.0b. I seem to not get as to what the parameters --genome-read-mismatches and --read-mismatches actually do.

I don't remember what --genome-read-mismatches parameter stood for, as its deprecated in 2.0.5 and is therefore not anymore on the website. Do you know how to find explanations of older versions' parameter from their webpage? As with --read-mismatches, from tophat webpage

--read-mismatches  Final read alignments having more than these many mismatches are discarded. The default is 2.

However, even if I have this parameter as 2, at the end, I get reads with mismatches more than 2 (the NM:i flag has values more than 2).

Answer 1

[aniprywatne]

Hi, it might be that the NM flag is wrong - I just realized this, when validating my tophat output with picard. When I'm validating it with the reference, I get message like this:

ERROR: Record 154910, Read name HWI-ST778:87:C0NMWACXX:3:1303:6497:61941, NM tag (nucleotide differences) in file [6] does not match reality [1]

I run Tophat with --read-mismatches=4 and --genome-read-mismatches=4, understanding that the latter is the number of mismatches allowed when the read is mapped against the genome, because it was unmapped against the transcriptome (where the --read-mismatches argument was used).

I also encountered another issue: I run tophat with --no-discordant --no-mixed , but I have many reads with flag 8 (mate unmapped).

[Arun]

Thanks for your feedback. I hadn't checked the validity of the NM flag. Let me check it and report back.