Analysis of biological networks through kernelized score functions

Several problems, such as disease gene prioritization, drug repositiong and protein function prediction can be modeled at system level as node label ranking problems on graphs. Most of the methods proposed in literature adopt local or global learning strategies to rank nodes or predict edges of the resulting graph.

We proposed a novel transductive approach that integrates both global and local learning strategies through graph kernels and simple functions able to learn the local features of each node of the graph (Valentini et al. 2016).

These methods have been successfully applied to the analysis of biomolecular networks for the prediction of protein functions (Jiang et al. 2016, Mesiti et al. 2012, 2014), for disease gene prioritization (Re and Valentini, 2012, Valentini et al. 2014) and drug repositioning (Re and Valentini 2012, 2013).

We also recently proposed novel semi-supervised method named Patient-Net (P-Net), based on graph kernels and score functions to analyze “patient graphs” instead of “biomolecular graphs”, as usual in the context of the Network Medicine. Biomolecular profiles of patients are modeled in a graph-structured space that represents gene expression relationships between patients. Then a kernel-based semi-supervised transductive algorithm is applied to the graph to explore the overall topology of the graph and to predict the phenotype/clinical outcome of patients. P-Net also provides interpretable models that can be easily visualized to gain clues about the relationships between patients, and to formulate hypotheses about their stratification (Gliozzo et al. 2020).

Publications

J. Gliozzo, P. Perlasca, M. Mesiti, E. Casiraghi, V. Vallacchi, E. Vergani, M. Frasca, G. Grossi, A. Petrini, M. Re, A. Paccanaro and G. Valentini. Network modeling of patients’ biomolecular profiles for clinical phenotype/outcome prediction. Scientific Reports, Nature Publishing, 2020.

G. Valentini, G. Armano, M. Frasca, J. Lin, M. Mesiti and M. Re. RANKS: a flexible tool for node label ranking and classification in biological networks. Bioinformatics, Oxford University Press 32(18), 2016.

Y. Jiang, T. Oron, W. Clark, A. Bankapur, D. D’Andrea, R. Lepore, C. Funk, I. Kahanda, K. Verspoor, A. Ben-Hur and Others. An expanded evaluation of protein function prediction methods shows an improvement in accuracy. Genome biology, BioMed Central 17(1), 2016.

M. Re, M. Mesiti and G. Valentini. An automated pipeline for multi-species protein function prediction from the UniProt Knowledgebase. Automated Function Prediction SIG-ISMB, 2014.

M. Re, M. Mesiti and G. Valentini. A fast ranking algorithm for predicting gene functions in biomolecular networks. IEEE/ACM Transactions on Computational Biology and Bioinformatics (TCBB), IEEE Computer Society Press 9(6), 2012.

M. Re and G. Valentini. Network-based drug ranking and repositioning with respect to DrugBank therapeutic categories. IEEE/ACM Transactions on Computational Biology and Bioinformatics, IEEE 10(6), 2013.

M. Re and G. Valentini. Large scale ranking and repositioning of drugs with respect to DrugBank therapeutic categories. International Symposium on Bioinformatics Research and Applications, 2012.

G. Valentini, A. Paccanaro, H. Caniza, A. Romero and M. Re. An extensive analysis of disease-gene associations using network integration and fast kernel-based gene prioritization methods. Artificial Intelligence in Medicine, Elsevier 61(2), 2014.

M. Re and G. Valentini. Cancer module genes ranking using kernelized score functions. BMC Bioinformatics, BioMed Central 13(14), 2012.

M. Re and G. Valentini. Random walking on functional interaction networks to rank genes involved in cancer. IFIP International Conference on Artificial Intelligence Applications and Innovations, 2012.

M. Re and G. Valentini. Genes prioritization with respect to Cancer Gene Modules using functional interaction network data.. NETTAB 2011 Workshop on Clinical Bioinformatics, 2011.