Hi @VerheulJulia,
It seems like 260-217 is getting around 30% passed the filtering set and then almost all of the reads are merging from there. Although that is not ideal it is probably the best you can get with this quality.
As for running single-end (just the R1 sequences) that could help get through the filtering steps because the forward read is usually better quality. I am not sure what you mean by merging in this step. Since you are doing just the R1 sequences there is no merging step, because you are not merging your forward and reverse reads. This is definitely a valid way to continue if you are concerned that you are losing too many reads.
Would you mind making a new question about the your classifier under user support. It is a little off topic and it would be helpful to answer these questions separately.
Thanks!
