MolGenie is developing and applying proprietary machine learning technologies to design novel small molecule compounds as new drug candidates to modulate protein folding and interactions. Thus, our PROtein Folding INDucers (PROFIND) may either stabilize or inhibit protein-protein interactions (e.g. SOS1-Ras interaction, attenuating MAPK signalling in oncology) or destabilize certain protein folds (e.g. reducing intracellular viral E7 protein load by proteolysis – PromptDegraderTM targets Intrinsically Disordered Regions (IDRs) of proteins, inducing their degradation, and is not reliant on one E3 ligase).
In particular, we are unlocking new chemistry spaces for drug discovery by selecting entirely new scaffolds with so far unknown tricyclic ring systems that are connected to and can be generated by established chemical reaction transforms. Using these rules and based on smart data extraction of proprietary structure-property-relationship datasets and 3D protein structures we design and synthesize molecules that realize desired functions, providing the foundation for a broad intellectual property estate.
Our compound design process includes 3 principal design stages:
First, pharmacophoric fingerprints are generated for a desired biological activity profile by 2 possible mechanisms:
– the protein fold and binding pocket is identified and a complementary imprint is calculated, holding properties, such as charge, polarizability, etc.
– alternatively, known structure-activity relationships (SAR) from known biologically active molecules are used to create flexible 2D or 3D pharmacophoric fingerprints.Second, molecules that satisfy the needs of the binding pocket or the SAR pharmacophores are designed – this includes the selection of intellectual property generating chemistries and building blocks from large libraries of chemical reaction transforms.
Third, molecules that have a good predicted bioavailability, solubility and managed metabolic stability are selected from the space of molecules designed in the second step. Furthermore, a broad screening for potential side and off-target effects is performed, excluding molecules with unwanted properties.
We have selected several high-value “hot” drug targets, aiming at the generation intellectual property on drug-like and selective small molecules that exhibit a clear structure-activity relationship and have the promise of good tolerability and oral bioavailability. These projects are conducted in collaboration with Actyon Discovery Inc., suitable contract research organizations and university biology laboratories specializing on disease models of interest.