Inspiration
Remember when we were all locked in our homes, bored out of our minds? Let's be honest, life would have been so much better if the COVID vaccine was discovered sooner. Drug synergy synergizes compounds to create more effective treatments for illnesses. Even when illnesses already have proficient solutions, synergizing compounds can create even more effective treatments and save lives. Drug synergy can also discover effective treatments for illnesses that don't already have proper treatments.
What it does
Compound Synergy computationally proposes over 13,000 possible combinations for drug synergy. Currently, searching for synergetic drugs is an arduous process and requires significant trial and error. By proposing synergetic drugs and ordering them by their likelihood of being effective, we save significant time when finding new medical compounds and help create more accurate predictions when testing new medicine.
How we built it
A library of 13,191 drugs and 47,728 proteins were obtained from the CANDO (Computational Analysis of Novel Drug Opportunities). Protein docking analysis (via the CANDOCK platform) was then performed on the set of drugs using the library of proteins. Each drug was then assigned a 'signature' which is a list of scores obtained from the docking analysis for each protein. A signature for EGT was also calculated using the .mol file of the compound and then compared (via the similar compounds method in the CANDO platform) with all other signatures to find the most structurally similar drugs. We allow users to enter drugs and illnesses to search for possible synergetic drug combinations. We use Flask webserver to run the application and connect the back-end with the front-end.
Challenges we ran into
Since we are dealing with an extremely large amount of possible drug combinations, processing a large amount of data was challenging. Furthermore, we spent a lot of time working on the front end so we could present the findings of our algorithms.
Accomplishments that we're proud of
We are proud of being able to use various metrics to actually assess how likely two drugs will be synergetic. In most current methods, this process takes significant trial and error and we are able to help reduce this time in an efficient computational manner.
What we learned
We learned a lot about writing code in an efficient manner and a lot about advanced applications of computer science in biology.
What's next for Compound Synergy
We are in the process of incorporating machine learning approaches, using both Machine Learning Classifiers and Deep Learning architectures, to provide scientists with further information on the potential of two drugs being synergetic.
Log in or sign up for Devpost to join the conversation.