Workshop:CAFCW20: Sixth Computational Approaches for Cancer Workshop
Authors: Benjamin Madej (Frederick National Laboratory for Cancer Research, Accelerating Therapeutics for Opportunities in Medicine (ATOM) Consortium); Neha Murad and Kishore Pasikanti (GlaxoSmithKline); Amanda Minnich (Lawrence Livermore National Laboratory); and Juliet McComas, Sabrina Crouch, Joseph Polli, and Andrew Weber (GlaxoSmithKline)
Abstract: The drug discovery process has been described as a large-scale multi-parameter optimization problem to find new chemicals to treat diseases. Not only must a new compound show efficacy to improve a disease state, a compound must also fit numerous criteria to become a new drug. A compound’s pharmacokinetics (PK) and toxicity properties are equally important and are even more difficult to predict. Many compounds in pre-clinical and clinical trials are discontinued due to toxic effects in animals and humans.
Given the considerable time and money investments in drug discovery projects, it is crucial to address PK and toxicity problems as soon as possible in the discovery pipeline. The Accelerating Therapeutics for Opportunities in Medicine (ATOM) consortium strives to tackle these problems in the drug discovery process by combining cancer informatics and high-performance computing approaches . ATOM has developed a range of PK machine learning (ML) models to predict PK and toxicity properties early in compound development . Through robust ML and mechanistic PK models, ATOM aims to provide early warning for PK and toxicity problems.