AlphaFold 2 is not good at predicting antibody binding, a point which DeepMind the developer of AlphaFold 2 made in the AlphaFold Multimer paper. But AlphaFold 3 has improved antibody binding predictions -- still not very good. In the AlphaFold 3 paper figure 1a it shows the success rate in a small benchmark predicting antibody binding went from about 30% in AlphaFold 2 to 60% in AlphaFold 3. So you should use AlphaFold 3 if you want to make antibody binding predictions. The ChimeraX AlphaFold prediction capability uses AlphaFold 2. Google has not released the source code for AlphaFold 3 so only Google's server can run AlphaFold 3 predictions.

On Nov 10, 2024, at 1:38 PM, karan kohli via ChimeraX-users <chimerax-users@cgl.ucsf.edu> wrote:

Hi,

Thank you for developing Chimera X, a tool that can help novices like myself to make in silico predictions.

I am new to the world of silico protein interaction prediction.

I am looking to use Chimera X to predict interaction between the antibody clones our company generated and the target protein.

This should be doable with Chimera X right. I

I am not sure how to do it.

I am trying to follow the below video., But I am struggling to enter my ab sequence as it contains two chains (heavy and light).

https://www.youtube.com/watch?v=TMcjEecFHaI

I would really appreciate some help.

Thank you,
Karan

_______________________________________________
ChimeraX-users mailing list -- chimerax-users@cgl.ucsf.edu
To unsubscribe send an email to chimerax-users-leave@cgl.ucsf.edu
Archives: https://mail.cgl.ucsf.edu/mailman/archives/list/chimerax-users@cgl.ucsf.edu/