Big Data and Statistical Genetics

The increase in the amount of data collected every day, in many disciplines, asks for new statistical strategies to handle it and to evaluate the research hypothesis timely. When dealing with high dimensional data, low-rank approximation based methods, such as singular value decomposition and principal component analysis are of great interest. In many of these data sets some variables and/or in individuals are more important and should be given higher importance/weight in the analysis. Therefore, algorithms for weighted low-rank approximations must be used. In this seminar I will be talking about big data and statistical genetics, two of the major hot topics in statistics nowadays. In particular, I will present some results on weighted singular interaction and QTL detection, in the context of plant genetics. Applications are made to observed and simulated data and the results compared with standard methods.