One of the most challenging problems arising in many scientific contexts is modelling data characterised by a huge number of variables interacting with each other in some complex and unknown pattern. Often, the sample size considered in the analysis is small compared to the number of variables. The development of new statistical tools proposed to analyse these data is therefore crucial in contemporary research development. In the research field of drug discovery, we deal with very high-dimensional and complex problems. More specifically, lead molecule optimization concerns the identification of molecules with required properties and the set of variables that affect these properties is extremely large. In this systems, the exploration of the experimental space entails experimentation that should to be limited since it requires high investments of resources. The construction of efficient experimental designs can contribute substantially to obtain valid and accurate experimental results at a minimum cost. Several studies have focused on strategies to design experiments inspired by evolution and the information gathered from statistical models in particular when the experimentation is conducted to search for an optimal value. One of the great benefits of these model-based evolutionary designs is that they are very flexible as they can be tailored to the problem under study. In addressing the lead molecule optimization from a statistical perspective, the task is to detect the set of relevant variables able to predict the desired properties of molecules. In this seminar we provide a strategy to reduce the dimensionality of the experimental space with an application to a drug discovery process.
Valentina Mameli is currently a postdoctoral research fellow at the Department of Environmental Sciences, Informatics and Statistics, Ca' Foscari University of Venice. She joined the European Centre for Living Technology in April 2016. She is working on a research project aiming to develop statistical models for reducing the dimensionality of chemical datasets. She obtained a Masters' Degree in Mathematics in 2008 at University of Cagliari and obtained a PhD in Matemathics and Scientific Computing at the same university in 2012. Previously, she worked as a postdoctoral research fellow in Statistics at University of Padova and at the University of Cagliari. In 2010 she was a visiting PhD student at the Statistical Laboratory, Centre for Mathematical Sciences, University of Cambridge (UK).