DocumentRX-6232
Issued
Read4 min

Humanitas exits stealth with LSU DeepDrug license and 104-program book

Humanitas launched at PharmSci 360 with the LSU DeepDrug license, 104 programs, SM-19 assets on hold, and a million-record EHR validation for its InteractIQ interaction checker.

TAG K-5499 · 739 words on the filing

After six years in stealth, Humanitas bets on hybrid AI for drug discovery
After six years in stealth, Humanitas bets on hybrid AI for drug discoveryJMD1 / pixabay

Summary of findings

  • Humanitas launched publicly on Oct. 27 at PharmSci 360 after six years in stealth, listing 104 programs across 19 asset classes
  • DeepDrug now returns 10-20 candidates per target, down from about 1,000 in 2020
  • SM-19 (400 mg imatinib + 100 mg mebendazole) cut viral load by up to 97% in cell and animal studies; trials are on hold pending acquisitions
  • InteractIQ validation targets one million LCMC Health records against EHR interaction alerts
  • 11 de novo antibiotic candidates, built with UBC's Neglected Global Diseases Initiative, advance to cell and animal testing

Humanitas Global Health emerged from six years in stealth on Oct. 27 at AAPS PharmSci 360 in New Orleans, announcing a commercialization agreement for Louisiana State University's DeepDrug platform and a portfolio its website counts at 104 programs across 19 asset classes and three companies. Founder and CEO Chris Galliano framed the debut as a prelude to a major funding round and called the company "less of a startup and more of a roll-up."

What does the hybrid architecture actually do?

DeepDrug pairs a large language model with what Humanitas calls a hypergraph that maps protein-protein, protein-drug and drug-drug interactions. "We use LLM as an orchestration layer, but we need another aspect of that to do more fine-grain analytics based on measurements," Galliano said. The graph keeps predictions traceable — traceability he says must "hold up in a regulatory package," a claim he says resonated with major pharma clients.

The lineage runs to 2015, when Galliano began working with Supratik Mukhopadhyay, the LSU professor who led the DeepDrug team. Both come from model-based systems engineering.

Chief Development Officer Frank Cerasoli — formerly of Pfizer, Roche and Sandoz, and part of the Contrave launch team — described the design loop: DeepDrug bolts together pharmacophores into candidate molecules, then iterates on absorption, pharmacokinetics and toxicology while preserving activity. "You kind of end up with some kind of Frankenstein molecule... that then gets revised, iterated over and over and over again," he said. Revised candidates run through an internal toxicity screener before cell and animal studies.

Why 10 to 20 candidates beats 1,000

The platform returned roughly 1,000 candidates per target in 2020; the current version produces 10 to 20. Galliano contrasted that with an LLM-driven screen that cuts a million compounds to 1,000 hits: "you still got to go wet lab on those 1,000, and that's a lot of cycle time, a lot of investment."

An antibiotic model built on a century of penicillin chemistry yielded 11 de novo candidates through the toxicity screener. The team will advance them to cell and animal testing, then match the strongest three against published critical-need lists, with sepsis high on the list. Development runs with the University of British Columbia's Neglected Global Diseases Initiative. The LSU team reached the IBM Watson AI XPRIZE semifinals in January 2020, one of 10 teams left from 147 that entered in 2016.

The COVID-19 detour and SM-19

Retooled for antiviral targets, the team mapped human proteins the virus depends on, narrowed them to about 25 targets and ranked approved drugs against entry, fusion and replication. "Human proteins would not change, whereas the spike was constantly changing," Galliano said. The result, SM-19, pairs 400 mg imatinib with 100 mg mebendazole — neither indicated for COVID-19 alone or combined. LSU reported SM-19 cut viral load by up to 97% in cell and animal studies. The project ran 13 months to the May 2021 FDA clearance for a first human study; European studies began Aug. 10, 2021, and Skymount received UK approval for a placebo-controlled trial in mild to moderate COVID-19 in February 2022. Both approvals remain valid; trials sit on hold pending Humanitas's acquisition activities.

InteractIQ and the EHR validation

The toxicity screener scaled from single drugs to combinations of 25 or more, computing ADME post-metabolism — "it's a different ball game" once a drug is metabolized, Galliano said. Humanitas packages the tool as InteractIQ, a point-of-care interaction checker for pharmacies and hospitals that grades patients on a zero-to-100 health score and recommends "the best, safest drug within that class." Medicare patients, he noted, often see two to seven specialists and take seven to 17 drugs daily. Cerasoli's charge against EHR alerts: "They miss quite a few, and those that they find are pretty much already known by the practitioner."

The retrospective validation uses LCMC Health records with LSU Health as collaborator. "Our goal, as the protocol is written, is to find a million records, and then from there, assess the interactions and the efficiency of that process inside the electronic health record versus the efficiency and the outcome from InteractIQ," Cerasoli said. Results will drive a prospective study.

Galliano would welcome a head-to-head with frontier models — "an Olympics... toe to toe with both of them. I would welcome it any time, anywhere."

via globenewswire.com (Original)

Share this article:

More from Olivia Hart

Olivia Hart

Show full bio

Senior reporter covering money, funding and market shifts around global pharmaceutical industry at PharmInsider.

22 articles

Linked filings

  1. K-1215
  2. E-6328
  3. E-9238
  4. V-9816
  5. K-4629

« Previous filingNext filing »