Ligand- and receptor-based approaches are combined in a virtual screening workflow that is tailored and validated for your project. A hit can be further optimized in order to improve its properties and/or to circumvent patent protection. We apply , an artificial intelligence approach for the optimization of structures according to multiple objectives (activity, selectivity, solubility, ADME …) based on genetic algorithms.
Understanding the activity of molecules based on their structure is key for optimization and validation of potential ligands in the drug development process. We are able to determine the most reliable binding poses and link protein-ligand interactions with structure activity relationship (SAR). We can complement the docking studies with alternative ligand-based studies: predictive pharmacophore- and/or molecular-field-based models which are suitable for SAR interpretation.
Our services in MOA are made in two steps: firstly, a determination of targets combing similarity searches and bioactivities using propietary software. Secondly, target specific analyses based on receptor dockings and pharmacophore information are applied.
Structure- and Ligand-Based Design of Small Molecules
We have developed several technologies specialized for designing small molecules. The input for these technologies depends on the information available: known active ligands or receptor conformation structures. We use a combination of different technologies (developed in-house and/or by third parties) in order to find the optimal molecules within the designed library. We apply machine learning algorithms that allow the search for optimal molecules according to a simultaneous multi-parameter evaluation within a large chemical library, including fragments.
Customized molecular modeling software migration to the cloud. We offer the full migration of commercial and/or in-house software so that customers can exploit the advantages of on-demand supercomputing.
We apply comparative modeling to build three-dimensional protein structures in the desired conformation. We can carry out molecular dynamic studies of biomolecules and their complexes with ligands, parameterizing atoms/molecules if necessary. We can study the stability of your molecules and estimate variations of free energy of multiple processes such as ligand-protein binding and mutations. We have excellent hybrid computational resources in order to execute your project in the best possible price quality range.
We predict ADME/Tox properties for a large group of molecules, using internal developed models that can be adapted to your molecules using propietary QSAR software and data mining. Applying machine learning methodologies, we build your own model for your specific molecular family, including PLS, SVM, and other methodologies, as well as a varied set of descriptors, ranging from 2D fingerprints to 3D molecular fields. State of the art data distribution, validation protocols, and applicability domain definition are incorporated to provide you with most reliable model possible within your data.
We can build relational and non-relational databases of biochemical data in order to organize your it, enable searches on it and generate prediction models with your data. We have experience in merging heterogeneous databases, creating special tools for fast cross-searching and the use of natural language when retrieving equivalent information from different source databases.