I am using zip pdb files in batchs of about 30-60 pdb AB complexes submitted to the PRODIGY web with EXCELLENT and reproducible affinity predictions. PRODIGY is helping me to select those few complexes with maximal affinities (lower Kcal) among many other complexes randomnly generated by RFdiffusion AI. THANK YOU !
My questions now are:
-
PRODIGY web. Could the Table appearing in the screen summarizing all the results be sorted by binding affinity?
-
PRODIGY local succesfully running in windows 10. How can I submit the zips? just like *.zip instead of *.pdb? & how can I submit code to obtain the summarizing screen wonderful Table also in local?
Thank you for your attention
Julio
Dear Julio
My questions now are:
- PRODIGY web. Could the Table appearing in the screen summarizing all the results be sorted by binding affinity?
This is going to be in place in a future update of the prodigy server.
You can already see it at: https://rascar.science.uu.nl/new/prodigy
- PRODIGY local succesfully running in windows 10. How can I submit the zips? just like *.zip instead of *.pdb? & how can I submit code to obtain the summarizing screen wonderful Table also in local?
The standalone code does not do that unfortunately. You would need to write a wrapper script to process your ensemble of PDB files.
You might open an issue with a new feature request on the prodigy GitHub repo.
Thank you for your responses Alexandre !
I visit your recomended new web site. However I was unable to submit a zip test. I could not fill out the A and B chains when submitting one of the previously PRODIGY runned zips. It says the A and B are not in contact, but the old web resolved the same complexes correctly. It may be still in development. Look forward to test it.
In the meanwhile, I developed a python script to run local PRODIGY using any open PyMol file as the Python-interpreter. The now called prodigybath.py works in windows 10 / 64.
In a few minutes runs the local PRODIGY for 60 pdb A B files from RFdiffusion list without any zipping. It generates an ordered txt list of the A-B interactions on kcal/mol. It may also be used to generate pmls and IC lists as in your PRODIGY web but I had not yet tested the appropriated subprocess.run – comands.
Although still needs some on-going fine-tunning, such a general py script may interest many other PRODIGY/window potential users…
If you want, I would really appreciatte your suggested changes if you or someone in yur team could take a look into it
Surely still needs many improvements …
Thank you for your attention
sincerely
julio
Thanks for the feedback.
However I was unable to submit a zip test. I could not fill out the A and B chains when submitting one of the previously PRODIGY runned zips. It says the A and B are not in contact
Could you please double-check if all the molecules inside this zip file are actually in contact (less than 5A apart) and that all of them have chain A and B?
And for inputting multiple molecules in one go that is indeed an useful feature to be added. Could you please open an issue here → github.com/haddocking/prodigy detailing an use case?
dear Honoratory
I checked again with another zip with 128 pdb files which as the first one was successfully submitted and processed before to the previous PRODIGY web.
I enclosed some of the individual files and their zip. The same zip was submitted before and a few minutes ago
The files resulted from a RFdiffusion run.
Each file contains a unique 60-mer protein target and 30-mer RFdiffusion designs predicting different sequences and conformations.
The RFdiffusion/alphafold program does not provide any relative aproximations of the interactions on kcal/mol.
Most of the random design are useless and need to detect those predicted the highest affinities (lower Kcal/mol)
As I mentioned before, the previous PRODIGY web beautifully processed all those files predicting kcal ranges, pml, pdbs, etc 
Although aproximated, PRODIGY allowed me to quicky select the best ligands according to the lowest kcal/mol and continue improvement efforts
Hope these may help
sincerely
julio
(Attachment all_pdb.zip is missing)
(attachments)
design0_n0.pdb (56.8 KB)
design31_n2.pdb (55.2 KB)
1 Like