Our research aims to analyse biomedical data efficiently, in particular we develop new methods to mining biological networks, integrate heterogeneous data, analyse omics, reconstruct pangenomes, analyse genomes haplotype-aware and to classify patients. We use theory coming from machine learning, data science, mathematics and graph theory.
We develop new computational approaches to investigate genetic variation effects. In particular, our research focuses on investigating potential effects of personal genetic variants on epigenetic factors, regulating chromatin structure and genic expression, by analyzing and integrating different NGS data. We implement methods and websites to predict targets for CRISPR-Cas9, with particular emphasis on assessing the potential effects of individual genetic variation on targets. The methods consider both single-nucleotide variants (SNVs) and indels, accounts for bona fide haplotypes, accepts spacer:protospacer mismatches and bulges, and is suitable for personal genome analyses. We also investigate the information content of biological sequences, from genomes to pangenomes, by means of alignment-free and/or reference-free methods based on information theory. The main goals are the identification of informative k-mers contained in genomes or pangenomes for evolutionary studies with application to biomarker discovery.
Through the analysis of multimodal bulk and single cell data we are interested in finding biomarkers for the early detection of neurodegenerative diseases in particular Alzheimer. We investigate the effects of air pollution in the development of Alzheimer and the biological mechanism that regulates such effects. We develop innovative clustering algorithms for spatial transcriptomics combining the information of stained tissue images with high-throughput spatially resolved RNA sequencing and expression values. Our methodologies have been also applied in immunology and cancer.
We develop efficient network based machine learning algorithms to classify patients or disease states integrating heterogeneous data, from imaging to multi-omics. We develop tools for integrating in comprehensive knowledge graphs the current understanding of biological phenomena and linked medical metadata. We develop algorithms to mine biomedical networks and in particular to find in exact or approximate ways substructures. Network algorithms are also used for understanding the role of non-coding RNAs, from small to long ones, in the regulation of cell functions. Investigating their relationships with the other molecules is a key factor for understanding their impact in the cell regulatome.