SignalP fails with >10,000 sequences

[embatty]

SignalP has a limit of 10,000 sequences and fails with the error message:
# Sequence limit reached: Max 10000 sequences are allowed

If you give Prokka a sequence with more than 10,000 coding sequences the SignalP step fails silently and it reports 0 signal peptides.

[tseemann]

Clearly your bacterium is too big (or fragmented).
I admit I don't run with signalp much these days.

What behaviour should prokka take if it finda 10,000 genes?
Should it die instead of running signalp?

[embatty]

Yes, these are really fragmented assemblies which is why they have a ridiculous number of genes... It would be neat if Prokka caught the SignalP error and gave a warning but continued the run, then I know it failed to run rather than having 0 signal peptides in my genome as a real result.

[tseemann]

@embatty I now check if no. seq > 10,000 and don't run signalp in that case. not the best solution but at least doesn't kill your job.

other option is to simply not use --gram option.