Is it methodologically appropriate to use fastp only for adapter detection and then perform the actual adapter trimming using Trimmomatic?
In other words, is it acceptable to identify the adapter sequences with fastp first and subsequently provide those sequences to Trimmomatic for precise adapter removal, or would it be more appropriate to use a single tool for both detection and trimming?
I don’t see any reason why that wouldn’t be valid, but!
First I would do both the things you suggested (finding adapters with fastp and then 1. remove them with fastp and 2. remove them with trimmomatic). Then benchmark the biological results -they should be the same more or less; if they are not, go with the tool that makes more sense. If they are similar, go with the fastest tool (time benchmark). But I guess the overhead of switching from 1 tool, to 2 tools probably leans the scale towards just using fastp.
TLDR; Benchmark results quality and execution times and go with what makes more sense
Now if you want a list and examples of various adapter removal and merge tools already in nf-core to use for your cases, have a look into the tools offered by this subworkflow: