[chimerax-users] Only retrieving the first predicted model in AlphaFold?

Tom Goddard goddard at sonic.net
Thu Sep 22 11:01:20 PDT 2022


Hi Yao-Chi,

AlphaFold by default predicts 5 models using 5 neural nets that were trained with different protocols.  There is no way to know which of the 5 will give the highest scoring model without computing all of them.  ChimeraX and the web browser ColabFold notebook

	https://colab.research.google.com/github/sokrypton/ColabFold/blob/main/AlphaFold2.ipynb <https://colab.research.google.com/github/sokrypton/ColabFold/blob/main/AlphaFold2.ipynb>

don't offer the ability to choose one of the 5 neural networks.  But you can make your own ColabFold or AlphaFold installation on your own computer and edit the code to do that.  That is a lot of work to set up, needs Linux and ideally a nice GPU.

Energy minimization removes atom clashes between neighboring residues.  Whether you need it depends on how you intend to use the predicted model.  The default in ChimeraX daily builds is to not energy minimize, but the default in ChimeraX 1.4 is to energy minimize.  You can set that option using the Options button in the ChimeraX AlphaFold panel.

	Tom


  
> On Sep 21, 2022, at 7:09 PM, Yao-Chi Chen via ChimeraX-users <chimerax-users at cgl.ucsf.edu> wrote:
> 
> Hi: 
> By ChimeraX, I am using AlphaFold to predict several protein-protein complexes. These complexes totally comprised about 600-800 amino acids.
> 
> To have the predicted results as soon as possible, I would like to ask: 
> 1: Is there a way to predict only one model(the best one)?
> 2: what's the advantage of using the Energy-minimized structures?
>     Could I just not use it?
> 
>  Thanks for your help.
> 
> 
> backy
> _______________________________________________
> ChimeraX-users mailing list
> ChimeraX-users at cgl.ucsf.edu
> Manage subscription:
> https://www.rbvi.ucsf.edu/mailman/listinfo/chimerax-users

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.rbvi.ucsf.edu/pipermail/chimerax-users/attachments/20220922/d6c900a8/attachment.html>


More information about the ChimeraX-users mailing list