[chimerax-users] AlphaFold prediction template structures

Hernando J Sosa hernando.sosa at einsteinmed.edu
Tue Sep 6 13:06:58 PDT 2022


Hi Tom,

Thank for the info, it was very useful to better understand what to expect from the Alphafold predictions.

By the way I managed to successfully run Alphafold for complexes of over 1200 residues total. I did not run into the download problem again, but the runs crashed a few times due to insufficient RAM. When this happened, I got a dialogue (from Google collab) asking if I wanted more RAM. After saying yes a rerun of the simulation, apparently from the beginning, completed successfully. I did upgrade my google account to collab pro.

Best

Hernando

___________________________________
Hernando Sosa
Dept. of Biochemistry
Albert Einstein College of Medicine
1300 Morris Park Av.
Bronx NY 10461
phone (718) 430-3456
emai:  hernando.sosa at einsteinmed.<mailto:hernando.sosa at einsteinmed.org>edu
___________________________________


________________________________
From: Tom Goddard <goddard at sonic.net>
Sent: Tuesday, September 6, 2022 3:04 PM
To: Hernando J Sosa <hernando.sosa at einsteinmed.edu>
Cc: ChimeraX Users Help <chimerax-users at cgl.ucsf.edu>
Subject: AlphaFold prediction template structures

CAUTION: This email comes from an external source; the attachments and/or links may compromise our secure environment. Do not open or click on suspicious emails. Please click on the “Phish Alert” button on the top right of the Outlook dashboard to report any suspicious emails.
Hi Hernando,

  When you run an AlphaFold prediction using the sequence of a structure you have open in ChimeraX, for instance, it shows "Sequence #1/A" to use the sequence of model #1 chain A, it only uses the sequence.  It does not use the atomic coordinates in any way.  AlphaFold can use structure templates but it is rather inflexible and finds them itself from the Protein Data Bank.  If you get a ChimeraX daily build and press the Options button in the AlphaFold panel it will show a checkbutton "Use PDB templates when predicting structures" (default false).  This checkbutton is not in ChimeraX 1.4.  You can turn that on an AlphaFold will find up to 4 templates from the Protein Data Bank.  AlphaFold does a really horrible job finding templates and it does not even report the templates it used.  Also in my tests trying to get AlphaFold two different structure conformations by running it with different template structures, it always found just one conformation.  It seems that the sequence alignment has much more influence on the final prediction than a template structure.

  Still if you want to try forcing AlphaFold to use your templates, the ColabFold notebook web server (not in ChimeraX) has an option to use custom templates.

https://colab.research.google.com/github/sokrypton/ColabFold/blob/main/AlphaFold2.ipynb<https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fcolab.research.google.com%2Fgithub%2Fsokrypton%2FColabFold%2Fblob%2Fmain%2FAlphaFold2.ipynb&data=05%7C01%7Chernando.sosa%40einsteinmed.edu%7Ce153e05c547042c7f20b08da903ab6d1%7C9c01f0fd65e040c089a82dfd51e62025%7C0%7C0%7C637980879479656375%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000%7C%7C%7C&sdata=m93LIau5esqM%2Feps%2BWRyPONIRCTDv6zrZD4gd%2BTVfHU%3D&reserved=0>

ChimeraX uses this same improved method (called ColabFold) of running AlphaFold but does not offer all of the options that the web server does.

Tom


On Sep 4, 2022, at 7:45 PM, Hernando J Sosa <hernando.sosa at einsteinmed.edu<mailto:hernando.sosa at einsteinmed.edu>> wrote:

 Hi Tom,

Not sure is I've been accepted yet in the ucsf-chimeraX list but perhaps I can ask you another Alphafold related question:.

When specifying the sequences to predict structures they van be pasted into the input sequence window, but they can also be taken from the structures/models already opened in chimera. In this second case does these 3D structures are used as templates or in any way  (other than the sequence)   in the Alphafold prediction?

Thanks

Hernando

___________________________________
Hernando Sosa
Dept. of Biochemistry
Albert Einstein College of Medicine
1300 Morris Park Av.
Bronx NY 10461
phone (718) 430-3456
emai:  hernando.sosa at einsteinmed.<mailto:hernando.sosa at einsteinmed.org>edu
___________________________________


________________________________
From: Hernando J Sosa <hernando.sosa at einsteinmed.edu<mailto:hernando.sosa at einsteinmed.edu>>
Sent: Sunday, September 4, 2022 10:59 AM
To: Tom Goddard <goddard at sonic.net<mailto:goddard at sonic.net>>
Subject: Re: [chimerax-users] AlphaFold

By the way I find a bit disconcerting the way ChimeraX arranges the windows.  For example, when the new window with the Alphafold run is hovered a bit over the ChimeraX window,  ChimeraX  grabs it and all windows inside ChimeraX gets rearranged.  It takes a while to put things back the way they were. Is there a button to reset the windows display ? Perhaps like other programs that have a view menu ?

Thanks again

Best

Hernando

___________________________________
Hernando Sosa
Dept. of Biochemistry
Albert Einstein College of Medicine
1300 Morris Park Av.
Bronx NY 10461
phone (718) 430-3456
emai:  hernando.sosa at einsteinmed.<mailto:hernando.sosa at einsteinmed.org>edu
___________________________________


________________________________
From: Tom Goddard <goddard at sonic.net<mailto:goddard at sonic.net>>
Sent: Friday, September 2, 2022 5:17 PM
To: Hernando J Sosa <hernando.sosa at einsteinmed.edu<mailto:hernando.sosa at einsteinmed.edu>>
Cc: chimerax-users at cgl.ucsf.edu<mailto:chimerax-users at cgl.ucsf.edu> <chimerax-users at cgl.ucsf.edu<mailto:chimerax-users at cgl.ucsf.edu>>
Subject: Re: [chimerax-users] AlphaFold

CAUTION: This email comes from an external source; the attachments and/or links may compromise our secure environment. Do not open or click on suspicious emails. Please click on the “Phish Alert” button on the top right of the Outlook dashboard to report any suspicious emails.
Hi Hernando,

  I have not seen that problem.  Does that always happen when you run an AlphaFold prediction?  What operating system are you on?  Mac?  I have seen Mac ask for permission to download from a web site the first time you download from that site.  It may be that the browser in ChimeraX showing the Alphafold run does not know how to ask.  But it doesn't make much sense because I do not observe this.  Another idea -- if your machine is sleeping when the job finishes maybe Google Colab gives up trying to download the results to your machine.  If it is sleeping you might want to set your machine not to sleep during the run.  This can also be important to avoid Google Colab dropping your job entirely due to inactivity.  I guess another possibility is that it could be a firewall problem where your network firewall blocks the download.  Since the download seems to succeed for most users, there seems to be something specific about your computer setup that causes the problem.

  At any rate, to get the results you should be able to click on Files in the left column of the AlphaFold google colab panel and among the files you should see results.zip which is what ChimeraX downloads when the job finishes.  There should be a vertical "..." if hover over that file that pops up a menu that allows you to download the file.  Try that.  ChimeraX is receiving the download and putting it in ~/Downloads/ChimeraX/AlphaFold/prediction_N.  It may also hang if there is some issue that prevents the downloads.  Possibly you could connect to Google Colab through a standard web browser and try downloading the results.zip file and maybe you would at least get an error that would help diagnose the problem.

Tom


On Sep 2, 2022, at 1:26 PM, Hernando J Sosa via ChimeraX-users <chimerax-users at cgl.ucsf.edu<mailto:chimerax-users at cgl.ucsf.edu>> wrote:

      Dear Chimera/ChimeraX,

I've been running Alphafold through UCSF ChimeraX and sometimes the procedure gets stuck indefinitely at the last step, downloading the results files (see screenshot).  Is this a known issue? is there a workaround ?

Thanks

Hernando

<image.png>
_______________________________________________
ChimeraX-users mailing list
ChimeraX-users at cgl.ucsf.edu<mailto:ChimeraX-users at cgl.ucsf.edu>
Manage subscription:
https://www.rbvi.ucsf.edu/mailman/listinfo/chimerax-users<https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fwww.rbvi.ucsf.edu%2Fmailman%2Flistinfo%2Fchimerax-users&data=05%7C01%7Chernando.sosa%40einsteinmed.edu%7Ce153e05c547042c7f20b08da903ab6d1%7C9c01f0fd65e040c089a82dfd51e62025%7C0%7C0%7C637980879479812593%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000%7C%7C%7C&sdata=bTrX3LU9nZDCF5SMTYNHJ0HBpj6i2xCPyqjDJb0yuI8%3D&reserved=0>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.rbvi.ucsf.edu/pipermail/chimerax-users/attachments/20220906/e14433d4/attachment-0001.html>


More information about the ChimeraX-users mailing list