Combining computational machine learning methods with phenotypic screening data can be used to enable the rapid assessment of compound libraries and selection of compounds for testing in vitro. I will describe several examples from my own experiences of collaborations with academic labs that have used this approach and led to exciting progress developing therapeutic treatments for rare and neglected diseases. Collaboration between Collaborative Drug Discovery, Inc. and Dr. Joel Freundlich and colleagues (Rutgers University). This SBIR grant lead to the combined computational and experimental identification of a series of triazines active against Mycobacterium tuberculosis which had laid dormant for over 40 years. This work is now part of a $26M NIAID center grant at Rutgers and the subject of additional SBIR grants submitted. (http://www.ncbi.nlm.nih.gov/pubmed/23521795) Collaboration between Collaborative Drug Discovery, Inc. and Dr. Jair Lage de Siqueira-Neto and colleagues (UCSD). This SBIR funded project identified several new molecules active in vitro and in vivo against Chagas Disease. One of which, pyronaridine, is a promising clinical candidate which is already approved in Europe as an antimalarial. A provisional patent has been filed and 2 additional grants have been submitted after this work. (http://www.ncbi.nlm.nih.gov/pubmed/26114876) Collaboration between Collaborations Pharmaceuticals, Inc. (http://www.collaborationspharma.com/) and Dr. Peter Madrid and colleagues (Stanford Research International) and Robert Davey and colleagues (University of Texas Medical Branch at Galveston). This project used machine learning models to quickly identify three in vitro nM inhibitors of the Ebola virus, several of which are approved for clinical use against other diseases. A provisional patent has been filed. These collaborations have developed new intellectual property, grants and papers. This approach combining computational and in vitro data can be replicated and scaled to move ideas rapidly to proof of concept with or without funding. As several of the molecules are already approved drugs with new uses, we also have a faster track to the clinic.

Using machine learning models based on phenotypic data to discover new molecules for neglected diseases

251st National Meeting (2016)

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