Batch coloring co-folded proteins and outputting PNG

Question

[dthorbur]

I'm new to using protein visualization tools and struggling to batch process my files.

I currently have hundreds of co-folded proteins using Alphafold Multimer, but it's taking too long to go through them all and manually coloring the two chains. And in the near future I will have thousands more. Is there an way I can batch color and export the PDB files through software like Chimera X or PyMol?

Ideally, I would be able to export a png/jpg with the proteins from different angles. Something like the default orientation the file is opened, then spin 90° on any axis, then 90° on the same axis, and again to get a complete view of the protein.

Ultimately, I am hoping each figure would look something like this:

Answer 1

[Mensur Dlakic]

These are all fairly standard PyMol operations, so I am sure it is possible. However, I doubt than there is a script out there to do exactly what you want, so you may need to learn some PyMol programming.

• https://pymolwiki.org/index.php/Main_Page
• https://pymolwiki.org/index.php/Advanced_Coloring

I do question your overall strategy, though. I think we can tell when the complex interface doesn't look right. My guess is that the one you showed above would be in that category. However, it is very difficult to tell that each complex that looks good to the eye is actually meaningful. Rather than making an image of every single multimer folded by AlphaFold, you may be better served to devise the pipeline that will check the binding energy and interface surface area of each complex. That should give you a more reasonable number of complexes that you could visualize manually.

[dthorbur]

Thanks. It was a lot easier than I had anticipated, so my mistake for posting so quickly without trying.

The issues we are having is dealing with a large number of spurious hits that look like they have a lot of contacts, so we were trying to identify new thresholds and metrics to use. Contact area is a good idea, ipTM, and pDockQ, are all also being added. But we still need to visualise a large number of cofolded proteins just so we can identify these thresholds. But this is an altogether different issue than what I've posted about.

Answer 2

[Jiyao Wang]

You can use iCn3D Pyhton script to output the image: https://github.com/ncbi/icn3d/blob/master/icn3dpython/icn3d_url/batch_export_png.py. You just need to replace the iCn3D URL in the script.

If your input are custom PDBs, you can store the PDBs in a server and access it using iCn3D, e.g., https://www.ncbi.nlm.nih.gov/Structure/icn3d/full.html?type=pdb&url=https://storage.googleapis.com/membranome-assets/pdb_files/proteins/FCG2A_HUMAN.pdb.

To rotate the structure, you can click the menu "View > Rotate > Rotate 90", then click the menu "File > Share Link" to get the full URL, e.g., https://www.ncbi.nlm.nih.gov/Structure/icn3d/full.html?afmem=off&mmdbafid=P69905&bu=1&date=20230716&v=3.27.0&command=load mmdbaf1 P69905 | parameters &mmdbafid=P69905&bu=1; rotate x|||{"factor":"1.000","mouseChange":{"x":"0.000","y":"0.000"},"quaternion":{"_x":"-0.6601","_y":"0.000","_z":"-0.2534","_w":"0.7071"}}.