Skip to main content

Table 1 Features and descriptors breakdown for protein, binding pocker, mutation and ligand representation

From: PSnpBind-ML: predicting the effect of binding site mutations on protein-ligand binding affinity

Protein, binding site and mutation features (202 features)

Protein (60 features)

Sequence-Order-Coupling Number (SOCN) descriptor

Binding site (9 features)

Secondary structure (4 features)

Accessible surface area (1 feature)

Buried and exposed residues (3 features)

Binding pocket volume (1 feature)

Mutation (133 features)

Secondary structure information (2 features)

Cysteine, glycine, and proline mutations (3 features)

Amino acid groups changes (7 features)

Properties of mutated AA and its surrounding (96 features)

Solvent Accessible Area (ASA) change (1 feature)

Phi and Psi dihedral angles (2 features)

FoldX energy terms (22 features)

Ligand features (54 features)

Descriptor category

Descriptors/Fingerprints

0D Descriptors

Molecular weight (1 feature)

Aromatic atoms count (1 feature)

Aromatic bonds count (1 feature)

All atoms count (1 feature)

N atoms count (1 feature)

O atoms count (1 feature)

Electronegativity (1 feature)

1D Descriptors

Number of Hydrogen donors (1 feature)

Number of Hydrogen acceptors (1 feature)

Number of rotatable bonds (1 feature)

Number of violations of Lipinski’s rule (1 feature)

Basic groups count (1 feature)

XlogP (1 feature)

AlogP (3 features)

JPLogP (1 feature)

2D Descriptors (Topological)

BCUT eigenvalue based descriptor (6 features)

Topological polar surface area (1 feature)

Fractional polar surface area (1 feature)

Small rings count of sizes 3-9 atoms (9 features)

Vertex adjacency information (1 feature)

Carbon connectivity types (9 features)

Atomic polarizabilities descriptor (1 feature)

3D Descriptors (Geometric)

Van der Waals Volume (1 feature)

Solvent accessible surface area (1 feature)

Momentum of inertia (6 features)

Radius of gyration (1 feature)