Research in Computer Science

Evolutionary Algorithms: Intelligent Pharma has a very active research line working on evolutionary algorithms. The encapsulation of genetic algorithms via web-services is adressed. Thereby, every newly starting application, which needs an optimization step, contacts the web-service running the genetic algorithms. This allows one to perform all the calculation in tailored hardware customized to run genetic algorithms. The distribution of software using web services helps amongst other in reducing the calculation time.

Computer scientists also research Gene Expression Programming (GEP). GEP allows one to infer with the outcome of mathematic analytical expressions. Hence, we can infer with the mathematical analytical expression of the behaviour of certain molecular properties, such as molecular volume, which changes due to the flexibility and energy of the system. The obtained information is analyzed by using machine learning tools allowing Intelligent Pharma to perform the QSAR of the future, i.e. not only based on constant molecular descriptors.

Machine Learning: Machine learning tools such as neural networks, support vector machines, etc. are classic approaches to solve several problems in computational chemistry. As in case of QSAR, for many computational chemistry problems the classic machine learning tools cannot attain better prediction taxes than 70-85%. At Intelligent Pharma, we have exhaustively studied the causes of failure, letting us conclude that one problem is the use of constant molecular descriptors, resulting from the treatment of molecules as constant and rigid.

Molecules are three-dimensional flexible entities. When determining molecular descriptors, this has to be taken into account for achieving higher success rates. The main inconvenience is that the classic machine learning tools are neither prepared nor designed to "learn to understand" and therefore use variable descriptors. Intelligent Pharma conducts research on new machine learning possibilities, mainly based on support vector machines, in order to enable the program to learn to treat functional data.

Distributed-, Grid- and Cloud-Computing: Intelligent Pharma is also researching ways to parallelize their proper developed computational technologies. The development, mainly based on grid-computing, is done in collaboration with the Barcelona SuperComputing Center-Centro Nacional de Supercomputación. In addition, Intelligent Pharma is also exploring the viability of using cloud-computing to apply its technologies to more scalable computing environments. Research is conducted on Intelligent Pharma´s supercomputer HYDRA as well as on the MareNostrum, the most potent supercomputer of Spain.