Bern, March 02, 2026
The University of Bern and Space Peptides Launch PPB3: Advanced Deep Learning Platform for Drug Target Prediction
BERN, SWITZERLAND – Researchers at the University of Bern’s Department of Chemistry, Biochemistry and Pharmaceutical Sciences, in partnership with Space Peptides Ltd., today announced the launch of Polypharmacology Browser PPB3 (https://ppb3.gdb.tools/). This web-based deep learning tool predicts biological targets of small molecules with unprecedented speed and breadth, and is now being commercialized by Space Peptides for pharmaceutical industry applications.
Unprecedented Scale and Scope
PPB3 leverages deep neural networks trained on ChEMBL 34, encompassing 2.5 million interactions between 1.19 million molecules and 7,546 biological targets. Unlike previous tools limited to single proteins, PPB3 covers:
- Single proteins (61%)
- Cell lines (19%)
- Organisms (14%)
- Protein complexes and families
This comprehensive approach enables early identification of off-target effects and toxicities—critical for reducing drug attrition.
Superior Performance
The platform employs a deep neural network converting molecular fingerprints into target predictions in under one second. Seven fingerprint options are available, including an 8192-bit fused fingerprint for enhanced accuracy. Case studies confirm PPB3 successfully identifies both primary targets and off-target effects.
Commercial Partnership
Space Peptides Ltd., which provided financial support for development, now commercializes PPB3 technology for pharmaceutical industry applications, offering customized implementations for drug discovery programs.
Availability
PPB3 is freely available at https://ppb3.gdb.tools/. Code and trained models are accessible via GitHub and Zenodo.
The underlying chemical database can be visualized and explored in chemical space using augmented reality, offering an immersive approach to navigate molecular relationships and target interactions.
Publication reference:
Polypharmacology Browser PPB3: A Web-Based Deep Learning Tool for Target Prediction Using ChEMBL Data, Maedeh Darsaraee, Sacha Javor, Jean-Louis Reymond, Journal of Chemical Information and Modeling, 2026, doi.org/10.1021/acs.jcim.6c00299
Media Contact:
Prof. Jean-Louis Reymond PhD, The University of Bern | jean-louis.reymond@unibe.ch
About Space Peptides
Space Peptides is a specialized Contract Research Organization (CRO) and Contract Development and Manufacturing Organization (CDMO) dedicated to advanced peptide drug discovery solutions and innovative peptide synthesis services. Building on its academic heritage from the University of Bern/Switzerland the company now integrates computational tools like PPB3 into its service portfolio. Space Peptides offers customized CRO support for early-stage target validation and hit identification, as well as scalable CDMO services for pre-clinical and clinical peptide supply, all under full cGMP compliance. For more information, visit https://spacepeptides.com.