Zhanglab forum

I want to predict the structure of some protein sequences in batches. I want to download QUARK to run locally, but I don't see the download link. Would you please give me a path to download the QUARK programs ?

Dear user,
you can download the C-QUARK package at https://zhanglab.dcmb.med.umich.edu/C-Q ... nload.html
Best
ITteam

Dear ITtem,
Thank you very much for your quick reply.
I downloaded C-QUARK severer days ago, and the also downloaded the large C-QUARK-library. By running C-QUARK I found it would take quite a while for the calculation of a single protein sequence using 2 cpus. Given I have more than 1k sequences to calculate, I meant to run a quick "ab initio simulation" by setting "-pipe QE", and I found it also take >4 hours for a testing using the example seq.fasta. Under "-pipe QE", I found C-QUARK to run Deep MAS, PSI-BLAST and so on. Does that mean C-QUARK in "ab initio simulation" model also let the sequence to query against the libraries? Would you please to give a suggestion to deal with the 1k sequences running task?


Best wishes,

Xiao

Hi Xiao,

The C-QUARK indeed need over 4 hours to get a protein, some even more longer time. QE is full ab initio pipeline without any templates. it runs DeepMSA and PSIBLAST for sequence homologous collection, where later will be used for contact prediction by deep-learning, those steps are not related structure template collection, so it is still ab initio. For the simulation step, you can reduce the time by "-hours" parameters. But notice this can also reduce the performance.

Best
ITteam

BTW, C-QUARK does not support multi-cpu, that means if you have two cpus, then you can run tow proteins at the same time, then each job will occupy one cpu. if you have multi-core or a cluster, 1K sequence can run parallel