I'm trying to do pan-genome analysis using BPGA and I face this error:
This error didn't show up in the other data set where all the .gbk is downloaded from NCBI. This .gbk file is our own data and has not been published in NCBI yet. There's of course protein data in it, there's this CDS qualifier "translation", but somehow BPGA didn't detect this.
The difference between our data and NCBI data is that most of the NCBI .gbk files are annotated with PGAP. On the other hand, our .gbk is annotated with Prokka (have tried it in DFAST and KBase). NCBI .gbk files that are annotated with PGAP have this CDS qualifier "protein_id" in them. Besides that, I don't think there's any difference.
Does the BPGA script counts the protein number based on the "protein_id" qualifier? How should I solve this? I can't annotate the .fasta file with PGAP because it's so resource-intensive. I believe that, if Prokka is installed locally, there's this "--compliant" option that can integrate the "protein_id" qualifier into the output .gbk file. I have tried to install Prokka locally, but it didn't run well too, I encountered another error for that and all the solutions didn't work for me.
Any opinion on this issue? Thanks for reading my question.
EDIT: I tried to insert one dummy "protein_id" on my Prokka-annotated .gbk files and it run perfectly without errors.
But when I tried to insert dummy protein_id on my DFAST-annotated .gbk files, it won't run and faced another error. What is this "protein_id" really defined by the script?