Link Prediction in Multipartite Graphs With Application to Drug Repositioning Studies

Area of Science:

• Computational biology
• Graph theory
• Drug discovery

Background:

• Developing new ethical drugs is time-consuming and costly, exceeding a billion dollars per drug.
• Drug repositioning offers a cost-effective alternative, with increasing reliance on in silico predictions.
• Existing graph-based methods struggle with datasets involving more than two data types.

Purpose of the Study:

• To introduce an innovative graph theoretical technique for imputing potential links in multipartite graphs.
• To address the limitations of current analytics for complex, multi-type biological datasets.

Main Methods:

• Developed a novel graph theoretical technique to impute missing edges in arbitrary multipartite graphs.
• Applied the method to five tripartite graphs, each containing disease, drug, and gene product sets.
• Interpartite edges represented known interactions or associations.

Main Results:

• Successfully demonstrated the utility of the new imputation method on tripartite graphs.
• Provided evidence supporting the reliability of the imputed edges.
• The technique enhances the analysis of complex biological interactions.

Conclusions:

• The introduced graph theoretical technique offers an effective solution for analyzing multi-type biological data.
• This method can advance drug repositioning efforts by improving the prediction of drug-target interactions.
• The approach has the potential to accelerate the identification of novel therapeutic applications for existing drugs.