Mining the Characteristic Interaction Patterns on Protein–Protein Binding Interfaces

Abstract

Protein-protein interactions are observed in various biological processes. They are important for understanding the underlying molecular mechanisms and can be potential targets for developing small-molecule regulators of such processes. Previous studies suggest that certain residues on protein-protein binding interfaces are "hot spots". As an extension to this concept, we have developed a residue-based method to identify the characteristic interaction patterns (CIPs) on protein-protein binding interfaces, in which each pattern is a cluster of four contacting residues. Systematic analysis was conducted on a nonredundant set of 1,222 protein-protein binding interfaces selected out of the entire Protein Data Bank. Favored interaction patterns across different protein-protein binding interfaces were retrieved by considering both geometrical and chemical conservations. As demonstrated on two test tests, our method was able to predict hot spot residues on protein-protein binding interfaces with good recall scores and acceptable precision scores. By analyzing the function annotations and the evolutionary tree of the protein-protein complexes in our data set, we also observed that protein-protein interfaces sharing common characteristic interaction patterns are normally associated with identical or similar biological functions.

Data Mining Approach for Extraction of Useful Information About Biologically Active Compounds from Publications

“…Therefore, an increase of the general consistency is observed when a compound is analyzed under similar experimental conditions. However, too low SD values, even lower than expected experimental error (SD­(p K i ) = 1 (for details see ref )), might be caused by repeatedly published values, which can be found in the literature, even when references to earlier studies were included (and would have sufficed). Such cases make it difficult to analyze the real activity of compounds and should be taken into account in the process of data mining for drug discovery.…”

“…Many studies are dedicated to the integration of chemical and biological data. Different approaches have been proposed to establish drug–target–disease relationships, identify protein–protein interactions, , interpret associations between proteins and genes, annotate proteins as well as protein expression and disease mechanisms, perform gene ontology analysis, , search for associations between drugs and diseases; and extract data on the melting points of chemical compounds …”

Prediction of water positions in the binding sites of proteins based on collections of multi‐source heterogeneous atoms

“…The triplets for which each edge of the triangles was larger than the maximum range of the edge between corresponding atom pairs in the training set were not considered. Additionally, 7.0 Å (two times the maximum range of hydrogen bonds) was set as the maximum threshold for the atom pairs that were not in the training set. For the retained collections of multi‐source heterogeneous atoms of triplets, a preference factor (Li, Liu, Han, Li, & Wang, ) ( PF ) value was adopted to determine the probability of the presence of water molecules around them. The shape‐matching method was used to match triangles with similar shapes in the training set to the retained collections of multi‐source heterogeneous atoms of triplets. The two ratios and the sum of the lengths of the three edges of the triangle formed by the collection of multi‐source heterogeneous atoms of a triplet were calculated.…”

Mechanistic insights into TNFR1/MADD death domains in Alzheimer’s disease through conformational molecular dynamic analysis

“…The appropriate hydrogen/hydrophobic interactions among proteins are involved in multiple biological processes and other underlying molecular mechanisms in different diseases 39 , 40 . Furthermore, prior research data also showed that the stability of protein complexes depends upon the hydrogen bonds 41 .…”