fibos 2.2.2

Package to calculate Occluded Surfaces

---

editor_options: markdown: wrap: sentence

FIBOS

The Occluded Surface (OS) algorithm is a widely used approach for analyzing atomic packing in biomolecules. Here, we introduce fibos, an R and Python package that extends the OS methodology with enhancements. It integrates efficient Fortran code from the original OS implementation and introduces an innovation: the use of Fibonacci spirals for surface point distribution. This modification reduces anisotropy and ensures a more uniform and even distribution of surface dots, improving the accuracy of the algorithm.

R fibos version is available to install by CRAN.

Operating Systems

FIBOS was designed to be multiplatform and run on Linux, Windows and Mac.
Tested on:

• Linux: Ubuntu ($\geq$ 20.04)
• Windows: Windows 11
• Mac: MacOS 15.0.1

Python versions

Tested on: 3.10, 3.11, 3.12

Compilers to different Python Versions

• gfortran
• gcc

Main functions:

1. occluded_surface(pdb, method = "FIBOS", density_dots = 5.0): Implements the Occluded Surface algorithm, generating points, areas, and normals for each atom. As parameters it accepts a PDB id (or the path to a local PDB file) and a method selection — either the classic 'OS' or the default 'FIBOS'. The function returns the results as a table and creates a file named prot_PDBid.srf in the fibos_file directory.

2. osp(file): Implements the Occluded Surface Packing (OSP) metric for each residue. Accepts a path to an .srf file generated by occluded_surface as a parameter and returns the results as a table summarized by residue.

3. get_radii(): Returns the Van der Waals radii values employed in the surface occlusion calculations. These are the values used in each computation.

4. set_radii(radii_values): Enables customization of the radii values used in surface occlusion calculations. To specify new values, the function accepts a DataFrame as its parameter.

5. reset_radii(): Reverts all modifications made to the radii values, resetting them to their default settings.

Quickstart:

import fibos
import os

# Calculate FIBOS per atom and create .srf files in fibos_files folder
pdb_fibos = fibos.occluded_surface("1fib", method="FIBOS", density_dots = 5.0)

# Show first 3 rows of pdb_fibos table
print(pdb_fibos.head(3))

#                     ATOM NUMBER_POINTS   AREA RAYLENGTH DISTANCE
# 0  GLN 1@N___>HIS 3@NE2_             6  1.287     0.791     5.49
# 1  GLN 1@N___>HIS 3@CE1_             1  0.200     0.894     6.06
# 2  GLN 1@N___>HIS 3@CG__             1  0.160     0.991     6.27

# Calculate OSP metric per residue from .srf file in fibos_files folder
pdb_osp = fibos.osp(os.path.join("fibos_files","prot_1fib.srf"))

# Show first 3 rows of pdb_osp table
print(pdb_osp.head(3))

#    Resnum Resname     OS  os*[1-raylen]    OSP
# 0       1     GLN  36.81          21.94  0.157
# 1       2     ILE  49.33          36.13  0.317
# 2       3     HIS  64.14          43.17  0.335

A more complex example:

import fibos
import os
from Bio.PDB import PDBList
from concurrent.futures import ProcessPoolExecutor
from functools import partial

# Auxiliary function to calculate occluded surface
def occluded_surface_worker(pdb_path, method):
    return fibos.occluded_surface(pdb_path, method=method)

# Get PDB file from RCSB and put it into the PDB folder  
# Rename the file appropriately and return path to it 
# (i.e., PDB/prot_8rxn.ent -> PDB/8rxn.pdb)
def get_pdb(id, path="."):
    pdbl = PDBList()
    new_path = os.path.join(path, f"{id.lower()}.pdb")
    if not os.path.exists(new_path):
        original_path = pdbl.retrieve_pdb_file(id.lower(), pdir=path, file_format='pdb')
        os.rename(original_path, new_path)
    return new_path

if __name__ == "__main__":

    # source of PDB files
    pdb_folder = "PDB"

    # fibos folder output
    fibos_folder = "fibos_files"

    # Create PDB folder if it does not exist
    os.makedirs(pdb_folder, exist_ok=True)
    
    # PDB ids list
    pdb_ids = ["8RXN", "1ROP"]

    # Get PDB files from RCSB and put them into the PDB folder  
    pdb_paths = list(map(lambda pdb_id: get_pdb(pdb_id, path=pdb_folder), pdb_ids))
    print(pdb_paths)
    
    # Detect number of physical cores and update cores according to pdb_ids size
    ideal_cores = min(os.cpu_count(), len(pdb_ids))

    # Calculate in parallel FIBOS per PDBid 
    # Create .srf files in fibos_files folder
    # Return FIBOS tables in pdb_fibos list
    worker_with_params = partial(occluded_surface_worker, method="FIBOS", density_dots = 5.0)
    with ProcessPoolExecutor(max_workers=ideal_cores) as executor:
        pdb_fibos = list(executor.map(worker_with_params, pdb_paths))

    # Show first 3 rows of first pdb_fibos table
    print(pdb_fibos[0].head(3))
    
    # Prepare paths for the generated .srf files in folder fibos_files
    srf_paths = list(map(lambda pdb_id: os.path.join(fibos_folder, f"prot_{pdb_id.lower()}.srf"), pdb_ids))
    print(srf_paths)
    
    # Calculate OSP metric by residue
    # Return OSP tables in pdb_osp list
    pdb_osp = list(map(lambda srf_path: fibos.osp(srf_path), srf_paths))
    
    # Show first 3 rows of the first pdb_osp table
    print(pdb_osp[0].head(3))
    
# OBS: If you need to run this example in a Jupyter Notebook, move the 
# occluded_surface_worker function to a "fun.py" file and import it as 
# "from fun import occluded_surface_worker"

Case Study:

Here we show a case study (currently only in R), aiming to compare the packing density between experimentally determined structures and the same structures predicted by AlphaFold (AF).

Authors

• Carlos Silveira ([e-mail])
  Herson Soares ([e-mail])
  Institute of Technological Sciences,
  Federal University of Itajubá,
  Campus Itabira, Brazil.

• João Romanelli ([e-mail])
  Institute of Applied and Pure Sciences,
  Federal University of Itajubá,
  Campus Itabira, Brazil.

• Patrick Fleming ([e-mail])
  Thomas C. Jenkins Department of Biophysics,
  Johns Hopkins University,
  Baltimore, MD, USA

References

Fleming PJ, Richards FM. Protein packing: Dependence on protein size, secondary structure and amino acid composition. J Mol Biol 2000;299:487–98.

Pattabiraman N, Ward KB, Fleming PJ. Occluded molecular surface: Analysis of protein packing. J Mol Recognit 1995;8:334–44.