## Graph Theory to Repurpose Existing Drugs

Gabapentin is a medication marketed for controlling seizures.  However, PubMed declares that it can also be used to “relieve the pain of diabetic neuropathy (numbness or tingling due to nerve damage in people who have diabetes), and to treat and prevent hot flashes.” [1]  In other words, aside from the intended effects of the drug, it also has positive side effects that could treat other illnesses as well.

Obviously when the scientist were developing Gabapentin, they weren’t attempting to cure all human illnesses, just trying to control seizures.  So how does it happen that Gabapentin is now used to relieve diabetic neuropathy pain and hot flashes?  Traditionally, after prescribing a medication, doctors would observe the health of the patients and take notes on any possible correlated behavior.  They would then send these reports to be gathered.  If the manufacturer wants to declare that their medication can be used for another purpose, they would shift through these reports to find any side correlations then test these findings.  A tedious process.

With CureHunter, a drug discovery software released earlier this year, this process can be narrowed down to as little as 15 minutes.  No, it’s not magic, it’s math and graph theory.

According to the descriptions given, CureHunter is a data mining software that shifts through its numerous databases of reports to generate networks for its included drugs and diseases.  The article even compares the interactions between drugs and diseases in this network to friend interactions on Facebook.  For any drug, the software can create a graph of positive and negative feedback between the drug, the symptoms it treats, and the biological agents it uses.  With these graphs, anyone can query for certain symptoms to treat, and the software will spit out the most compatible available drug for the current case.  In other words, allocating an old drug new illnesses to treat. [2]

The graph theory used in these operations is obviously closely knit to the materials that we’ve covered in class.  We’ve talked about positive and negative edges, a variation used in this software for determining which symptoms can be treated by which drugs.  Whichever drug have the strongest links to the given symptoms will be recommended.  A scene reminiscent of suggesting friends on noted in lecture.

While most of our applications of graph theory have been social, this new software just goes to show that the lessons we learn in class can be fitting for circumstances outside the norm as well.