How to use AlphaFold3 to predict protein structures

Molecular modeling serves as an indispensable technique in biological research, allowing scientists to delve into the architecture and functionality of biomolecules. Among the various tools available for this purpose, Alphafold 3 stands out as a cutting-edge application. This advanced deep learning model is capable of predicting protein structures with remarkable precision. Here is the information on how to use AlphaFold3.

I. Website and Login

The server URL for AlphaFold3 is:

https://golgi.sandbox.google.com

II. Complex Structure Prediction

  1. Protein-Metal Ions

The supported metal ions include Mg, Zn, Cl, Ca, Na, Mn, K, Fe, Cu, Co, these ten metal ions. Take 4R59 as an example, which combines with the metal ion zinc: Enter the protein sequence, then click Add Entity, and select Ion

undefined

Click continue and preview, and after a while, you will get the results:

undefined

The color of the markings above indicates the confidence of the predicted structure, with blue being the best and orange being the worst. The error matrix on the right, the greener, the better the prediction.

undefined

When aligned the predicted structure with the crystal structure, and the RMSD obtained is 0.113, which is a very good result, and the metal sites basically coincide.

  1. Protein-Small Molecules

Unlike metal ions, select Ligand, the server only supports nineteen ligands, which are ADP, ATP, AMP, GTP, GDP, FAD, NAD, NAP, NDP, HEM, HEC, PLM, OLA, MYR, CIT, CLA, CHL, BCL, BCB.

undefined

  1. Protein-Nucleic Acids

DNA supports sequence input:

undefined

RNA also supports sequence input:

undefined

  1. For Non-Standard Amino Acids and Nucleic Acids

For processing non-standard amino acids and nucleic acids, click on the red part:

undefined

Then the following interface will appear, click on the amino acid to be modified:

undefined

Then, for example, the S here can be changed to NAG, and then click save. DNA has Reverse and Modification operations, RNA only has Modification.

III. Parameters

  1. Copies

undefined

For multimers, the structure is built by setting the copies here. For example, for a pentameric protein, setting the copies to 5 will result in:

undefined

  1. Seed

undefined

The seed is an initial seed, and different settings can result in different structures. For example, the following sets the seed to random, 1, 2, 3, and 100:

undefinedundefined

It can be seen that the obtained structural metal centers are different, and the position of nucleic acids also has a larger difference. However, the seed is not the larger the better. The results of 2 and 100 here are close, but 100 will take more time. Therefore, the specific settings should be based on the characteristics of the required structure.