Description
The challenge of an aging society with anticoagulant requirements of increasing complexity is a problem of great relevance in health improvement and encourages the search for new anticoagulant molecules. The objective of the project is to implement and apply an improved integrated computational-experimental strategy for the discovery of bioactive compounds that uses hybrid mathematical and artificial intelligence techniques on HPC architectures. The result will be the development of an improved drug discovery methodology that will cover any other drug discovery campaign, by using hybrid artificial intelligence and molecular modelling based on virtual screening methods on HPC architectures. Beneficiaries will be unhealthy patients that can only use heparin as anticoagulant. The project promoter Saint Anthony Catholic University of Murcia is an expert on Bioinformatics and High Performance Computing, the University of Alicante has expertise on Neural Networks and the donor partner, University of Iceland/ Nordic High Performance Center has expertise on Mathematical Artificial Intelligence Methods Development and High Performance Computing.
Summary of project results
The challenge of an ageing society with anticoagulant requirements of increasing complexity is a problem of great relevance. Heparin is widely used as activator of antihrombim, but incurs serious side effects. The main aim of the project was to implement and apply an improved integrate computational-experimental strategy for the discovery of bioactive compounds that used for the first time hybrid mathematical and artificial intelligence techniques on High Performance Computing architectures, in addition to advanced computational drug discovery methods already developed by the applicant. The project was divided in two main workpackages, developed consecutively: 1) Development and exploitation of improved VS methods. The aim was to develop improved VS workflow that will provide accurate hit predictions and will be experimentally characterized and analysed in Workpackage 2. It included: preparation of large compound databases; artificial intelligence based virtual screening; docking with BINDSURF; molecular dynamics. 2) Experimental characterization, analysis and optimization of VS hits. The aim was to test experimentally, both in vitro and in vivo, hits predicted in WP1 and their chemical optimizations. Further analysis of collected results delivered feedback for its utilization in WP1 in successive stages. This workpackage included: in vitro determination of activity and function; in vivo determination of activity and function; analysis of results; optimization of hits and compound synthesis; publications of results. During the project, participants attended and gave numerous seminars and prepared at least 15 papers, 18 entries were provided to conference proceedings, 11 talks were given in conferences, 3 posters were submitted and five PhD thesis were in preparation. Besides, several project proposals have been submitted to different funding schemes, in order to continue the joint research initiated within this project. All results, reflected in publications, conference contributions, thesis etc., will have an impact in the scientific community. Results and methodological advancements are available online in highly accesses scientific information sources in most of the cases. Most of the results obtained are contributing now to the development of methods for drug discovery and the promoter (UCAM) is progressing in the discovery of blood anticoagulants and other bioactive compounds.
Summary of bilateral results
Partners developed the project jointly throughout the whole implementation period. The UCAM (Catholic University of Murcia) managed the project at the technical and financial levels. Also worked extensively on the scientific tasks of the project. The UA (University of Alicante) group contributed to the project with their expertise and knowledge in Computational Intelligence and Computer Vision method to improve the traditional Virtual Screening methods developed by the UCAM group. The Nordic HPC of the University of Iceland worked very professionally during the whole project providing availability to supercomputing resources and support for using them in all the scientific objectives of the project. The group at the University of Iceland as international experts on Random Forest and other Machine Learning techniques, worked directly providing their experience and know-how to improve drug discovery techniques. As a result of the project, numerous publications have been prepared, PhD thesis improved, and conferences given. Partners expect to submit joint applications to various funding schemes in order to continue the joint research, such as H2020 programme of the European Union, the National Science Foundation at United States of America, or the National research funds in Spain, Iceland and Norway.