The usage of TriPepSVM about new taxon ids.

[xj7757417771]

Hello,
I recently used your TriPepSVM to predict RBPs, but the species used included A.thaliana and S.cerevisiae. The taxon ids were 3702 and 4932, respectivel through uniprot, but I got errors as follows:
Python: can't open file 'get_rbps_for_taxon.py': [Errno 2] No such file or directory
Fatal Error:
Failed to open the database file
Program halted !!
Rm: cannot remove ‘non-RBP/A//allData/RBP_3702.fasta.clstr’: No such file or directory

But on human species, it works fine.

The code I run is as follows:
./TriPepSVM.sh -i non-RBP/S.cerevisiae-non-RBP135 -o non-RBP/S/ -id 4932 -pos 1.8 -neg 0.2 -thr 0.68
I don't know where the problem is?

Looking forward to your reply, thank you.

Juan Xie

[bressin]

Hi Juan Xie,

Thank you for pointing this out. We fixed the bug. Please pull or download the new version.

However, you should be aware of two other "issues":

1. The class weights from the paper are optimized for human, salmonella and e.coli. We recommend you to use the inverse proportional feature weight (default) for a new taxon or you should optimize them by your own.

2. The taxonomy "4932" has just a few directly annotated RNA-binding proteins. Please try -r TRUE and/or a higher taxon order to enable classification.

Have fun!

Annkatrin Bressin