Artificial Intelligence for Drug Discovery
AI Platform to Discover Therapeutics for Substance Use Disorders
Collaborative effort on NIH Small Business Technology Transfer (STTR) Phase I grant from the National Institute on Drug Abuse (NIDA)
Using AI to Discover Novel Therapeutics for Addiction
Addiction and SUD arise from intricate networks of biological pathways. Many current therapies fail to address these overlapping mechanisms. Our project deploys powerful graph machine learning techniques on knowledge graphs — sophisticated structures connecting genes, proteins, drugs, pathways, and disease phenotypes — to:
AI Platform to Discover Therapeutics for Substance Use Disorders
Substance use disorders (SUDs) are a global burden in healthcare, with over 35 million people afflicted worldwide. Both those who suffer from SUDs, their family members and caretakers suffer from higher rates of suicide and mental health disorders. Unfortunately, medicine and science has not developed new treatment options for these populations, mostly due to the highly complex physiology and pharmacology of central nervous system (CNS)-related diseases.
In this proposal, we will use a hypothesis-based system to identify, validate and screen potential known and unknown targets to treat the behavioral phenotype of addiction. Recently, medications which target the glucagon-like peptide 1 (GLP-1) endogenous system and pathways have demonstrated a reduction in the phenotype of addiction. This system interacts with the dopaminergic mesolimbic system, which is strongly implicated in a reduction in ‘craving’ for addictive substances, including opiates, nicotine, alcohol and hedonistic over consumption.
We propose using artificial intelligence (AI)-based systems to interrogate these different pathways which all work within the reward-based addiction system. The overall premise and goals of this project is to leverage the combined expertise of the research team to develop: 1) a set of curated training datasets and machine learning (ML) models that accelerate key drug discovery tasks, 2) mechanistic targets for GLP-1 receptor agonist pathways which can be explored experimentally, and 3) preliminary hits for compounds that can modulate the group(s) of AI-recommended targets.
This will enhance our understanding of reward system physiology and pharmacology, as well as lay the foundation to develop an AI-based platform for drug repurposing using similar phenotype-based treatment options for CNS-related disease.
Graph Neural Network (GNN) models will be trained to identify: 1) possible targets which singly show a reduction in addictive behaviors, and 2) grouping of two or more targets that work together to reduce addictive phenotypes.
Top pathways and targets (differentially expressed genes or signaling pathways) will be validated using known pharmacological modulators of those targets. Ligand-based drug screening will also be used for pathway/target validation. A set of promising compounds will be identified that interact with possible targets of interest with high potency and selectivity.
Successful completion of this Phase I project will demonstrate feasibility of the ML models, validation and drug target screening using the compound libraries available in the Assay Development and Drug Repurposing Core at Michigan State University. In addition, this phase includes planning and development for Phase II, where we continue to explore, refine and validate targets as well as use pre-clinical model behavioral testing, confirming discovery targets.
The future goals of this project include development of these ML models into a platform that pharmaceutical companies can use to repurpose currently FDA-approved medication by symptom and/or behavioral phenotype.
This NIDA STTR project is a collaborative effort across a cross-disciplinary team of experts .
Principal Investigator
An Assistant Professor of Pharmacology at Michigan State University, Dr Lauver is leading this translational R&D project. His extensive training in pharmacology and drug discovery equips him to effectively lead this project. For over a decade, he has spearheaded various drug discovery programs in his laboratory for identifying new targets for therapeutic intervention and developing innovative treatment options for cardiovascular diseases. Dr Lauver has recently initiated new research endeavors aimed at enhancing preclinical drug safety analyses.
Expert Consultant
An Assistant Professor of Internal Medicine at University of Utah, Dr Wynne is serving as a subject matter expert vetting the open data sources and knowledge graphs generated by the team. With two decades of background in inflammation and pharmacology, as well as cardiovascular and renal physiology, she is perfectly suited to contribute to the research aims. Dr Wynne is a physiologist with a strong background in pharmacology and receptor biology. Her research program investigates the physiology of salt-sensitive hypertension to elucidate the mechanisms linking inflammation, salt and hypertension.
Callentis is a small research, development, and services company, focused on integrating the latest proven technologies into existing business processes to create growth and opportunity. With extensive background in engineering and computational science, our team offers software and hardware development for a variety of science and technology applications, and diagnostic support for data-driven decision processes.
Meet the NIDA STTR project leadership team.
ML Engineer & Project Manager
Dr. Kargah-Ostadi has over fifteen years of experience in applied statistics, data science, machine learning, and evolutionary computation. He has published over 20 peer-reviewed articles, and presented at over 20 conferences, workshops, and webinars. He holds a PhD in Civil Engineering and a doctoral minor in Computational Science from Penn State University. He is a registered Professional Engineer and a certified Project Management Professional.
AI/ML Specialist
Dr. Drach has over fifteen years of experience in practical AI/ML applications, applied biostatistics, biomedical and biochemical modeling, computational simulation, custom software and hardware development. He has published over 40 peer-reviewed articles, presented at over 20 international conferences, and gave 4 invited talks. He did his postdoctoral training in Computational Engineering and Sciences at the University of Texas at Austin. He holds a PhD in Mechanical Engineering from the University of New Hampshire.
© 2016-2025 Callentis Consulting Group, LLC. Running on ☕ from Austin, TX. All rights reserved.
