From a Nature paper to a global technology platform.
Innophore’s roots go back to the research labs of Graz, Austria. One idea has stayed the same the whole way: proteins matter, but binding sites matter even more.
The “Google for enzymes”
The team publishes the foundational Catalophore™ method in Nature Communications. Austrian and international media dub it a search engine for new enzyme functions. The same year it wins the international CPhI Pharma Award, sometimes called the “Pharma-Oscar,” along with the ÖGMBT research award.
Innophore GmbH is founded
Innophore spins off from the Austrian Centre for Industrial Biotechnology (acib) and the University of Graz, with support from Science Park Graz, and becomes a company in its own right: an enzyme discovery business already extending its technology toward recognizing drug side effects.
Into industrial biotechnology with new and novel enzymes
Early industry partnerships follow, including a collaboration with SignalChem, as the platform reaches beyond enzyme discovery toward drug-safety questions.
A pandemic stress test
Within hours of the SARS-CoV-2 genome going public, we model its main protease (Mpro) and flag it as a prime drug target, the same site antivirals like Paxlovid would later go after. Our models are downloaded tens of thousands of times, and we reach the finals of the JEDI “Billion Molecules against COVID-19” Grand Challenge.
virus.watch
With the AWS Diagnostic Development Initiative, we launch virus.watch to track emerging SARS-CoV-2 variants in near real time, backed by molecular-dynamics simulation.
Scaling with NVIDIA
We become one of the earliest adopters of NVIDIA BioNeMo and begin moving binding-site analysis onto GPUs. Industrial collaborations span detergents, materials and pharma, with partners including Henkel, Continental and Merck.
Into pharma with DrugSolver CavitomiX
We publish our drug-repurposing and off-target pipeline in Scientific Reports, recovering known drugs as SARS-CoV-2 inhibitors purely from binding-site similarity, independent of sequence or fold.
Folding the human proteome
Together with NVIDIA, we model the 3D structures of 42,042 human proteins, cutting whole-proteome modeling from over a year to about two weeks. The dataset is published in Scientific Data (Nature).
CavitOmiX goes public
Our NVIDIA partnership is announced at the J.P. Morgan Healthcare Conference and CES 2025. GPU acceleration takes the platform to five million off-target predictions per second, and we launch the CavitOmiX Copilot to help catch drug-safety risks before the clinic.
Proteins matter, but binding sites matter even more.
The conviction behind every chapter