Internal error xalip when setting cutoff level to -1

[matthiasblum]

Hello,

I am trying to reproduce ScanProsite results with pfscanV3. However, for some sequences, I am encountering an error when running pfscanV3 with -L -1 (to run the scan at a low confidence cut-off).

Getting data

$ wget ftp://ftp.expasy.org/databases/prosite/prosite.dat
$ wget https://rest.uniprot.org/uniprotkb/Q840Q1.fasta

Testing with Conda/Mamba

Create a new environment:

$ mamba create -n ps -c bioconda --quiet --yes pftools
Preparing transaction: ...working... done
Verifying transaction: ...working... done
Executing transaction: ...working... done

Activate it:

source activate ps

Run pfscanV3:

$ pfscanV3 --matrix-only -o 4 -L -1 prosite.dat Q840Q1.fasta > /dev/null
Error: Inconsistent alignment found in alignment 1 - no list produced.
       Alignement should be from 367 to 165!
Thread 3 : Internal error xalip reported no possible alignment for sequence 3(0) (nali=-1)!
>tr|Q840Q1|Q840Q1_STRGR DAGKc domain-containing protein OS=Streptomyces griseus subsp. griseus OX=67263 PE=4 SV=1
Segmentation fault (core dumped)

Testing with Docker

$ docker run --rm --quiet \
>     -v $PWD:/data \
>     sibswiss/pftools:3.2.12 \
>     pfscanV3 --matrix-only -o 4 -L -1 /data/prosite.dat /data/Q840Q1.fasta > /dev/null
Error: Inconsistent alignment found in alignment 1 - no list produced.
       Alignement should be from 367 to 165!
Thread 3 : Internal error xalip reported no possible alignment for sequence 3(0) (nali=-1)!
>tr|Q840Q1|Q840Q1_STRGR DAGKc domain-containing protein OS=Streptomyces griseus subsp. griseus OX=67263 PE=4 SV=1

Could you please have a look?

Thank you

---

Edit: I just realized this issue describes the same error as reported in #22.

[smoretti]

We can reproduce the issue with a compiled version of pftools3 from scratch. We will investigate it

In the meantime, could this workaround satisfy you? pfscanV3 --matrix-only -o 4 -L -1 prosite.dat Q840Q1.fasta || pfscanV3 --matrix-only -o 4 prosite.dat Q840Q1.fasta Run with -L -1 and if fails, re-run without it

[matthiasblum]

Thanks for the update. We (InterPro) currently run pfscanV3 or pfsearchV3 without -L -1, and I think it's better to keep it this way until a fix is found.