Our main research areas are bioinformatics and computational bioloy, especially computational statistics applied to the analysis of "high-throughput" data, mainly in cancer evolution. Our work ranges from the application of standard techniques to the development of new statistical approaches and their implementation as publicly available software, with special emphasis on the use of high performance computing.
Some of the problems we have worked on include patient classification and gene differential expression using microarray expression data, survival analysis with "omics" data, functional annotation of results from analysis of "omics" experiments and, segmentation of array CGH data to detect copy number changes in genomic DNA and to identify recurrent regions of alteration in groups of patients. Our current work involves the use of probabilistic graphical model to identify restrictions in the orer of accumulation of mutations and the prediction of trajectories of tumor evolution.
We elaborate on the current main research areas. There is plenty of cross-sectional data in publicly accessible data bases; at the same time, it is thought that many mutations in cancer are only possible if other mutations have taken place before. Different methods, known as "cancer progression models" (CPMs) have been developed to identify restrictions in the order of accumulation of mutations from these cross-sectional data. Our work focuses on understanding under which evolutionary and genetic scenarios CPMs are, or not, successful; for instance, what is the effect of reciprocal sign epistasis and different kinds of fitness landscapes on the performance of CPMs. Additionally, we are interested in extending these methods to estimate evolutionary predictability, obtain the probability of different paths of tumor evolution, and predict the probability of what is the next most abundant genotype in cancer progression conditioning on the current state of the tumor.