Our innovation and technologies

Chemical Space Big Data

The largest peptide database till date, mapped with molecular fingerprints to decode the space of peptides.

Key Technology

A portfolio of AI-augmented technologies and state-of-the-art algorithms (similar to technologies Google uses to identify handwriting), empower the evaluation of billions of molecules to design drugs faster and cheaper.

TMAP visualizes enormous data

GDBspace enumerates the world’s largest collection of virtual chemical molecules (>166 billion). Our proprietary TMAP platform organizes multiple existing libraries of molecules on a visual map. The resulting google map enables easy browsing of molecules and unlimited forms of combination beyond the boundaries of conventional discovery of novel molecules (e.g., lead finding in pharmaceuticals). TMAP offers intuitive human insights to empower smarter, faster and cheaper discovery predictions, virtually unlimited.



Selected Publications:

  • Daniel Probst et al., “Visualization of very large high-dimensional datasets as minimum spanning trees”, J. Cheminform., 2020.
  • Clémence Delalande et al., “Optimizing TRPM4 inhibitors in the MHFP6 chemical space”, Eur. J. Med. Chem., 2019.
  • Marion Poirier et al., “Identifying lysophosphatidic acid acyltransferase beta (LPAAT-beta) as the target of a nanomolar antiogenesis inhibitor from a phenotypic screen using the polypharmacology browser PPB2”, ChemMedChem, 2019.
  • Daniel Probst and Jean-Louis Reymond, “FUn: a framework for interactive visualizations of large, high-dimensional datasets on the web”, Bioinformatics, 2018.
  • Justus Bürgi et al., “Discovery of potent positive allosteric modulators of the α3β2 nicotine acetylcholine receptor by a chemical space walk in ChEMBL”, ACS Chem Neurosci, 2014.
  • Jean-Louis Reymond et al., “Exploring chemical space for drug discovery using the chemical universe database”, American Chemical Society Chemical Neuroscience, 2012.

PPBs – Artificial Intelligence (AI) based target prediction

By finding out if a newly identified bioactive molecule is closely related to molecules with documented bioactivity and therefore likely to interact with the corresponding biological targets (e.g., PPB2, the polypharmacology browser incorporating deep neural networks, naïve Bayers and/or k-nearest neighbors algorithm), one can perform on- and off-target predictions to de-risk your project, e.g.:

  • Target identification for phenotypic screening hits

  • Drug repurposing

  • Off-target risk assessment

Selected Publications:

  • Mahendra Awale and Jean-Louis Reymond, “The Polypharmacology Browser PPB2: Target Prediction Combining Nearest Neighbors with Machine Learning”, J. Chem. Inf. Model., 2018.

MQN - Scaffold hopping generates molecule ideas beyond the obvious

Our unique MQN fingerprint is the most performant molecular search available, handling billions of compounds per query. Compatible with the largest commercial (ZINC) and in silico generated databases (GDB). Used in ligand based virtual screening campaigns – this can be applied in early stage (hit/lead finding), but also in advanced projects if scaffold variations are required.

Selected Publications:

  • Philippe Schwaller et al., “Mapping the Space of Chemical Reactions using Attention-Based Neural Networks”, Nat. Mach. Intell., 2021.
  • Alice Capecchi et al., “One molecular fingerprint to rule them all: drugs, biomolecules, and the metabolome”, J. Cheminformatics, 2020.
  • Josep Arús-Pous et al., “SMILES-Based Deep Generative Scaffold Decorator for De-Novo Drug Design”, J. Cheminformatics, 2020.
  • Josep Arús-Pous et al., “Exploring Chemical Space with Machine Learning”, Chimia, 2019.