AlphaFold – run from ChimeraX

3D Protein structure prediction (3)

Previous posts (AlphaFold background, AlphaFold code) introduced AlphaFold and where the protein structure prediction could be installed, or run on the Colab cloud computing.

Colab or and Colab Pro

Whether using the Colab code detailed in the previous post as Jupyter Notebooks, or the method in ChimeraX below, it should be noted that the free Colab version has limitations that may prevent the prediction to complete. This would be particularly true for longer sequences and for Complex models. Therefore, Colab also provides a monthly subscription for US $10/month to access faster compute systems. See: colab.research.google.com

ChimeraX screenw
Menu: Tools > Structure Prediction > AlphaFold provides access to AlphaFold in Colab.

ChimeraX

A newer and perhaps easier possibility is to use the newly introduced tools within the molecular graphics and modeling software UCSF ChimeraX (basic ChimeraX tutorial.)

The new menu “Tools > Structure Prediction > AlphaFold” provides a simple interface that does not require the user to know how to use a Jupyter Notebook.

The recent changes have happened in September 2021 to provide this tool in ChimeraX version 1.3. (Download.) However, that version only supports monomeric predictions.

An update to compute multimeric “Complex” models will be provided in ChimeraX version 1.4 yet to be released, but already available in the “Daily Build” (Download.) The new AlphaFold “Complex” can model multimeric models with different sequences.

ChimeraX co-author Tom Goddard has reeleased a set of short (~15min) videos on how to use AlphaFold within ChimeraX. The most recent entries is shown below:

Running AlphaFold to Predict Protein Complexes from ChimeraX

https://www.youtube.com/embed/6lXeCPuTePs”

How to predict the structure of a photoreaction center complex of 3 proteins from ChimeraX using AlphaFold-Multimer. Requires a ChimeraX daily build newer than December 2, 2021 or version 1.4. (Download 1.4 or daily build) It is not in ChimeraX version 1.3. Web page describing steps

He also provided a written entry to follow along:

Running AlphaFold on Complexes from ChimeraX
Tom Goddard – December 7, 2021

More videos

Five more videos are also available on the ChimeraX channel.

Most show the ChimeraX tool to access AlphaFold prediction methods or pre-computed database.

  • Compare AlphaFold and Experimental Protein Structures in ChimeraX (8:24)
    • How to fetch a predicted protein structure from the AlphaFold database and compare it to an experimentally determined structure using the ChimeraX visualization software.
  • Predict a protein structure using AlphaFold within ChimeraX (7:56)
    • We run AlphaFold to predict the structure of the protein avidin (from chicken). We start the computation using ChimeraX (Sept 2021 version – Download) which runs it on Google Colab servers
  • Assembling AlphaFold protein models with ChimeraX (3:26)
    • We combine two AlphaFold protein structure predictions into an assembly using ChimeraX and find that its conformation does not match the experimental structure because the proteins are flexible. Requires September 2021 ChimeraX of newer (Download.)
  • Fit an AlphaFold database model to a cryoEM map (12:23)
    • We show how to start building an atomic model into a cryoEM map using an atomic structure from the AlphaFold database. We fit the atomic model into the map and adjust the position of one of its domains. The map is EMDB 30495 and is of a human dimer originally thought to be an mechano-sensitive ion channel, although the Nature publication speculates it is instead involved in lipid metabolism.
  • Run AlphaFold in ChimeraX for cryoEM model building (16:46)
    • We show how to run AlphaFold from ChimeraX to predict a protein structure from its sequence to produce an initial model for building into a 3 Angstrom resolution cryoEM map. We look at a protein that transports omega-3 fatty acids across the blood-brain barrier, EMDB 23883. Requires ChimeraX 1.3 or a daily build newer than September 2021 (Download.)