DIAMOND in reverse?

[PhilliVanilli]

Hi

We have a number of raw nanopore nucleotide reads of eg 300-1000bp in a big fastq file and would like to search for an 8 amino acid peptide in there. We don't know the underlying code of the peptide so basically, I was thinking of running DIAMOND but then instead of an NCBI protein database, just use a single peptide in the database. But I'm not a bioinformatician so I don't understand how DIAMOND works under the hood and whether this is at all possible or whether we can trick the program into allowing such a search. I've tried it and the software works but doesn't output anything. After all, the database peptide is so much smaller than the nanopore reads (8 AA vs .100-300 AA) . But maybe we can run DIAMOND in reverse ?

I also wonder how DIAMOND deals with the translation of raw nanopore data, I would think that can't work very well at Q20 with lots of homopolymer error induced frameshifts.

If you could let us know if anything is possible or even give us a clue as to what tools could work in this regard (e.g. a tool that can translate noisy nanopore reads), that'd be already greatly appreciated.

Thanks so much and thanks for providing this great software to the community!

[bbuchfink]

Diamond by default does not work well on very short sequences, but setting some parameters can help. I have given tips about it in the past, if you search the issues and discussions. You should use --min-orf 1 and can try with or without frameshift alignment mode (only way to deal with frameshifts). I'm not aware of any other tool for that purpose.