"Using AlphaFold2 to Predict the Conformations of Side Chains in Folded Proteins"

doi: 10.1101/2025.02.10.637534

Code to reproduce the CHI values reported in the manuscript:

Step 1: Convert AlphaFold (AF3) CIF files to PDB

AlphaFold3 currently outputs protein models in .cif format. To calculate side-chain χ (chi) angles, we first convert these CIF files to the standard .pdb format using PyMOL.

You can do this with the provided script:

convert_cif2pdb.py This script:

• Loads all .cif files in the working directory
• Converts them to .pdb format
• Saves them with the same base filename
• How to Use: Make sure to update the os.chdir(r"/working/directory") at the bottom of the script to your actual path. Then run script directly inside the PYMOL

Step 2: Compute Side-Chain Chi (χ) Angles for All Residues in Chain A

After converting .cif to .pdb files, this step computes χ1–χ4 angles for residues in chain A across all models (PDB files) using:

🔗 Chi_compute_multiple_pdbs.py

This script:

• Iterates over all .pdb files in a directory
• Computes χ angles using PyMOL dihedral calculations
• Outputs a .dat file for each structure with residue-level χ1–χ4 angles
• How to Use: Make sure to update the directory_path at the bottom of the script to your actual path. Then run script directly inside the PYMOL

Step 3: Merge All Chi Angle Files into a Single CSV for Analysis

Each .dat file generated in Step 2 corresponds to one PDB structure. This script aggregates all those into a single CSV file with one row per residue per structure:

🔗 convert_dat_csv.py

This script:

• Parses all *_chi_angles.dat files in a directory
• Extracts χ1–χ4 values for each residue
• Adds a column for the PDB ID (parsed from the filename)
• Merges all data into a unified CSV file for downstream analysis
• How to Use Update the paths at the bottom of the script: directory_path = "/path/to/folder/with/dat/files" output_csv_file = "/desired/output/path/AF3_chi_angles.csv"

python convert_dat_csv.py

Example files are uploaded in the **Example_AF3 for quick test