I am running since a long time now Mothur on my V4 sequenced samples. I like this program a lot especially the error correction feature of make.contigs. My primers do not perfectly align, like 50nt. Would it be a good thing to use maxee error check on those non-overlaping (or is it already being done?) small sequences to improve error discarding early in the process. Like if I run maxee=2 and deltaq=6, would it flag as N incorrect nulceotides in the csmall bits that do not overlap so the can be discarded?
Personally, for my V4, I first make my contigs using other software (mefit) that I think allows to a more complex handling of the problem of the lack of full overlapping… But still I discard 40% of the reads (all the non-overlapping for at least 50, plus those with bases Q<30 in any of both reads, or any ambiguity, or…).
In our testing deltaq is still much better than maxee. I’m not sure what you mean by the V4 primers “do not perfectly align, like 50 nt”. Do you mean that the two reads only overlap by 50 nt? If that’s the case then I wonder if you’re really sequencing the V4 region or if you’re not using the 2x250 nt chemistry.