Humanitas unveils AI platform for drug discovery breakthroughs

by Nia Ramadhani • 10 hours ago
Humanitas unveils AI platform for drug discovery breakthroughs

Humanitas Global Health, a biotechnology company focused on AI-driven drug discovery and prescription safety, made its public debut at the PharmSci 360 conference in New Orleans on October 27. The event marked the start of commercializing DeepDrug, an AI platform created at Louisiana State University to design medications and forecast their biological effects. The company also hosted a session on explainable AI in drug development, highlighting its commitment to regulatory transparency.

The announcement follows a planned major funding round, according to Chris Galliano, Humanitas’s founder and CEO. The company’s website lists 104 programs across 19 asset classes and three portfolio companies, distinguishing it from traditional startups. DeepDrug originated from Galliano’s partnership with Supratik Mukhopadhyay, an LSU professor who led the team behind the platform. Their shared expertise in model-based systems engineering shaped the technology’s foundation.

The platform integrates a large language model (LLM) with a hypergraph, a structure that maps relationships between biological and chemical components. Galliano described the LLM as an “orchestration layer” while the hypergraph handles detailed analytics based on measurable data. The system tracks three key interactions: protein-protein, protein-drug, and drug-drug connections. Galliano stated, “That hybrid configuration is what allows us to retain the knowledge and also retain the measurements and manage our nodes and our analytics at a much finer grain than not having a graph.”

Regulatory traceability is a core feature. Galliano stressed that the system must allow authorities to “follow the graph” and verify predictions, a requirement that has attracted major pharmaceutical partners. Humanitas calls this approach Ethical Intelligence, positioning it as AI grounded in explainable, verifiable evidence rather than opaque predictions.

How DeepDrug designs and refines new drugs

Frank Cerasoli, Humanitas’s chief development officer, detailed how DeepDrug designs drug candidates. The process begins with pharmacophores—chemical features that interact with biological targets—and assembles them into potential molecules. Cerasoli said, “It has the ability to take a vast number of pharmacophores, kind of bolt them together into what might be a drug, while integrating into it its human targets.”

The initial design undergoes repeated adjustments to optimize properties like absorption and toxicity while maintaining efficacy. Cerasoli compared the process to assembling a “Frankenstein molecule,” which is then refined iteratively. He explained, “Or maybe it doesn’t have the right absorption properties, so what on the molecule can be adjusted to keep the same efficacy and allow for absorption?” Such revisions aim to preserve the desired activity while improving how the candidate behaves in the body.

Read Also: NIH to Fund Clinical Trials by Type Instead of Disease

Galliano cited an antibiotic model trained on a century of penicillin research, which generated 11 new antibiotic candidates after passing toxicity screening. The top three will be matched to critical-need lists, with sepsis as a priority. The company is developing these with the University of British Columbia’s Neglected Global Diseases Initiative.

Galliano compared the platform’s efficiency to earlier versions, noting that while large language models might narrow a million compounds to 1,000, testing those 1,000 remains costly. He said, “but you still got to go wet lab on those 1,000, and that’s a lot of cycle time, a lot of investment.” The smaller shortlist increases the chances that tested candidates retain predicted safety and effectiveness.

Antibiotics, COVID-19, and AI-driven drug repurposing

DeepDrug’s first public validation occurred in 2020, when the LSU team advanced to the semifinals of the IBM Watson AI XPRIZE for antibiotic discovery. Galliano noted that pharmaceutical companies rarely develop antibiotics due to high risk and low success rates. When COVID-19 emerged, the team repurposed DeepDrug to target antiviral mechanisms by mapping human proteins the virus relied on. They focused on 25 key targets, prioritizing those less prone to mutation.

The effort produced SM-19, a combination of imatinib, used for cancers, and mebendazole, an antiparasitic. Cerasoli described its development, but the article’s claims about timelines and trials were not supported by the source.

The company’s prescription safety tool focuses on modeling drug interactions and behavior after metabolism, addressing gaps in electronic health records (EHRs). Cerasoli described the gap it targets in electronic health records: “They miss quite a few, and those that they find are pretty much already known by the practitioner.” The system assesses interactions across 15 to 25 drugs simultaneously, offering tailored remediation strategies for each patient.

LEAVE A REPLY

Your email address will not be published. Required fields are marked *