Giters

rdk / p2rank

P2Rank: Protein-ligand binding site prediction tool based on machine learning. Stand-alone command line program / Java library for predicting ligand binding pockets from protein structure.

Home Page:https://rdk.github.io/p2rank/

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Different results compared to PrankWeb?

AlexLam37 opened this issue · comments

commented

This is the same/related issue I've previously asked in the prankweb page however I thought it was useful to ask here as well.
cusbg/prankweb#142 (comment)

I am trying to use p2rank to predict pockets for a bunch of proteins but I noticed there are different results compared to the web server. I.e. if I go to the web server and use the "predicted structure" and input a uniprot ID, it seemingly gives different results compared to if I download the AF model from uniprot, and then run that on p2rank.

Similarly if I download the AF model and submit as custom structure to the web server, it gives same result as the standalone p2rank.

Is the web server over/underestimating pocket probabilities? Please see below for an example using A0A644EZV9

Using "predicted structure" on prankweb
image

Using "custom structure" on prankweb, same results as running standalone p2rank
image

commented

See the explanation at cusbg/prankweb#142 (comment) .