In silico Identification and Characterization of Protein-Ligand Binding Sites

Abstract

Protein-ligand binding site prediction methods aim to predict, from amino acid sequence, protein-ligand interactions, putative ligands, and ligand binding site residues using either sequence information, structural information, or a combination of both. In silico characterization of protein-ligand interactions has become extremely important to help determine a protein's functionality, as in vivo-based functional elucidation is unable to keep pace with the current growth of sequence databases. Additionally, in vitro biochemical functional elucidation is time-consuming, costly, and may not be feasible for large-scale analysis, such as drug discovery. Thus, in silico prediction of protein-ligand interactions must be utilized to aid in functional elucidation. Here, we briefly discuss protein function prediction, prediction of protein-ligand interactions, the Critical Assessment of Techniques for Protein Structure Prediction (CASP) and the Continuous Automated EvaluatiOn (CAMEO) competitions, along with their role in shaping the field. We also discuss, in detail, our cutting-edge web-server method, FunFOLD for the structurally informed prediction of protein-ligand interactions. Furthermore, we provide a step-by-step guide on using the FunFOLD web server and FunFOLD3 downloadable application, along with some real world examples, where the FunFOLD methods have been used to aid functional elucidation.

Determining protein similarity by comparing hydrophobic core structure

“…In summary, the Authors would like to point out that symbiotic collaboration between experimental structural biology and computational biology may benefit both disciplines. Indeed, the CASP community has recently reached out to CAPRI to facilitate joint studies into the function/structure relationship, acknowledging that proper identification of protein complexation and ligand binding sites may provide important clues regarding the protein’s biological role [ 56 , 57 , 58 , 59 ]. Application of the FOD model to p-p interfaces has revealed some new aspects of complex generation, including the presence of an independent “quasi-domain” comprised of fragments contributed by both chains participating in the complex [60] .…”

Proteins and Their Interacting Partners: An Introduction to Protein–Ligand Binding Site Prediction Methods

“…3. Most prediction servers, such as COACH [ 49 ] and FunFOLD [ 3 , 4 , 57 ], utilize in-house structure prediction pipelines to construct models for protein–ligand interaction predictions that may not always produce the best quality model for every target, which may result in over- and under-predicted protein–ligand binding sites. Nevertheless, despite these shortcomings, prediction methods are constantly under development and improvements can be gauged via the rigorous independent blind assessment scoring, described in Section 3 .…”

“…The majority of these methods predict putative protein–ligand binding sites and ligand binding site residues, while some methods additionally predict Enzyme Commission Numbers (EC) and Gene Ontology (GO) terms. We have developed a number of versions of a template-based method, called FunFOLD [ 3 , 4 , 57 ], which starts with a 3D model of the target protein predicted from sequence, for example using the IntFOLD server [ 63 , 64 ]. Each version of the algorithm has worked on the assumption that proteins with the same fold that bind to similar biologically relevant ligands are likely to have similar binding sites.…”

“…The algorithm outputs scores for five sequence- and structure-based features that are combined using a neural network, outputting predicted Binding-site Distance Test (BDT) [ 68 ] and Matthews Correlation Coefficient (MCC) [ 69 ] scores. The FunFOLD3 [ 57 ] pipeline additionally outputs a set of per-residue binding probability scores to comply with the CAMEO-LB format [ 65 ]. Furthermore, the FunFOLD3 method outputs a putative ligand binding site, putative ligand binding site residues, putative ligands that may bind to the target protein, along with predicted EC and GO [ 70 , 71 ] terms (see Section 4 ) for each target protein [ 3 , 4 ].…”

“…The majority of methods predict function based on Gene Ontology (GO) terms (which include: INGA [ 78 ], EFI-EST [ 79 ], SIFTER [ 80 ], GEO2Enrichr [ 81 ], PANNZER [ 82 ], and PILL [ 83 ]) with fewer utilizing EC numbers (EFI-EST [ 79 ] and DomSign [ 84 ]) for functional annotation. Furthermore, a number of structure-based methods for the prediction of protein–ligand binding sites have incorporated methods for predicting GO and EC terms, including COACH [ 49 ] and FunFOLD3 [ 3 , 4 , 57 ] (See Section 2.2.4 ). However, as these methods build 3D models as part of their prediction pipeline, they are somewhat more computationally intensive than the sequence-only methods.…”