It looks like the issue here is coming from the fact that most your reads aren't able to be merged - because a very high number of them are being retained post-filtering. Is there a reason that you're trimming your sequences? If they are EMP protocol, there is no need to trim your sequences - and with 150 bp reads, merging can be tricky since there isn't as much room for overlap as with 250 bp reads.
I'd recommend increasing your trunc-len parameters as well to increase the number of retained reads on the 3' end. You could remove the trunc-len param on your forward reads, as the median value remains pretty high throughout. I'd personally increase the trunc-len on your reverse reads to somewhere around 140 (which is about what you have in your most recent run) since the median quality score does seem to drop quite a bit after that, and the range in quality score is pretty large there as well.
If these adjustments don't produce a reasonable number of merged reads (I'd generally aim for at least greater than 50%) then you may just not be able to work with paired end reads in your data, and may just need to work with single end reads for your analysis (even though I realize that's not ideal).
I am trimming because I'm not using 'cut adapt' and because I found noisy the range of quality, nowO see I'm just wrong. I will repeat it changing parameters. Do I loss much using the single end approach?
Your total read length will be shorter (150 bp for your reads) since you aren't merging your forward and reverse reads - but if you're unable to merge them, that will be your best way forward.
forum!
