Finding safe, non-toxic, and specific molecules to address a given research question is the dream of every Merck scientist. We propose a foresight mechanism that compiles various evidence from literature, clinical trials, and real-world data that can accelerate and focus this search process. By analyz- ing these resources together computationally we can produce a map of each therapeutic molecule and its association with genes, proteins, cell type, cell line, RNA, chemical entities, diseases, model organisms, bioassays, and other data that can be incorporated into our property graph. Such a map can pro- vide grounds that we can further explore molecules and find the most specific to the question being asked. The MolecRank foresight mechanism provides a view of the feasibility of the under-development molecules in question, shorten some paths in discovery and development and rank, according to user-defined search facets, the most specific therapeutic molecules to scientists. MolecRank is a Machine Learning system to analyze these resources, extract and build a map (we refer to as a network) of molecules and their associated biological features. The network is further explored using means of connections from the molecules to whatever evidence of features found in the resources. We designed a network analysis algorithm that rank molecule, in a similar manner to that of Google ranking websites, but based on significantly different criteria.<br/>We establish our molecule ranking based on the following factors: (1) the specificity of the molecule, (2) the toxicity of a molecule, (3) the overall knowl- edge we assemble and aggregate about a molecule allows us to map a picture about its potential safety, efficacy, and risk of development. This weeds out all the non-relevant data to the questions asked. We have currently implemented a preliminary version of the work using a small subset of publications. The algorithm can be run and tested on Algorithmia. This information would be pushing scientists to investigate new drugs in the right direction. It shortens the lifecycle of the discovery, eliminates the chances of failure, and cut the cost significantly. We are extending the scope of the algorithm to integrate clinical-trials data and real-world evidence.

Therapeutic molecule foresight tool saving lives and saving billions of dollars

256th National Meeting (2018)

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.