Use of Drug Interaction Networks for Drug Repositioning
The knowledge of the exact mechanism behind the action of many pharmaceutical ingredients is often limited. If we don’t know quite how a drug works, it can be hard to predict what else it might be useful for. A drug can be approved for one use, but it may turn out to be incredibly useful for treating a different ailment altogether. This drug repurposing, or repositioning, has become a more and more common route for drug companies as it has become increasingly difficult to have new bioactive molecules approved for use by the FDA and other authorities. In fact in 2013, 20% of ‘new’ drugs were repositionings. This repurposing is traditionally done by trial and error informed by biological intuition and the hunches of medical professionals. The recent study of the network properties of drug-drug interactomes (DDIs) has sped up this process and given it a more quantitative grounding. DDIs are networks with a drug as each node and the edges denoting an interaction between drugs when used together.
The analysis of these interaction networks allows pharmaceutical researchers to, among other things: predict previously unknown drug interactions, assure the avoidance of adverse interactions of new drugs, extract information about the pharmaceutical properties of drug molecules. In the paper cited, the DDI is generated from a raw drug interaction network using clustering, or community forming, algorithms to create groups of highly similar drugs. The sort of clustering algorithms used were initially studied for the analysis of social networks. Using network derived qualities like the degree and closeness centrality of each node, the interaction potential of every drug is determined. The communities formed by clustering are labeled by the dominant common pharmacological property. In confirming these properties of the nodes within the labeled clusters, 63% were listed in drug databases and another 22% of the node properties were confirmed after more in depth cross referencing. The remaining 15% not explained were flagged as possible candidates for repositioning, as the drug quality not listed in any database may simply not have been investigated yet. Additional nodes that might be repurposed were those sitting on the boundary between certain communities. There is still a huge potential for further study in the network analysis of drugs, and it could have life saving effects for those waiting for new drugs to be released.
https://www.nature.com/articles/srep32745