Carlo Vittorio Cannistraci
Machine learning and complex-network science for personalized and systems biomedicine.
Previous and Current Research
The Biomedical Cybernetics Group adopts a transdisciplinary approach integrating information theory, machine learning and network science to investigate adaptive processes that characterize complex interacting systems at different scales, from molecules to organisms, in biology and medicine. This knowledge is leveraged to create novel and more efficient artificial intelligence algorithms; and to perform advanced analysis of patterns hidden in biomedical data, signals and images. Our theoretical effort is to translate advanced mathematical paradigms typically adopted in theoretical physics (like topology and manifold theory) to characterize many-body interactions in quantitative biology. We apply the theoretical frameworks we invent in the mission to develop computational tools for systems and network biology, personalized biomedicine and combinatorial drug therapy.
Plasticity phenomena – like remodelling, growth and evolution – modify the topology of complex systems, their internal state and their multidimensional representation in form of networks or high-dimensional datasets. Our theoretical mission is to elucidate the general rules and mechanisms that underlie this type of structural plasticity, which is at the basis of learning and memory processes in living organisms. In particular, we develop methods for topological analysis of self-adaptive and self-organizing learning systems such as protein interaction and epigenetic networks at the molecular level, and brain networks at the cellular level.
In neuroscience, we are interested in how the brain networks wire at synaptic and functional levels to modulate learning processes. And, on a molecular pathway scale, we seek to identify the network patterns that could suggest which broken functional-modules are responsible for memory aberrations in neurodegenerative diseases. Since general paradigms of regeneration and degeneration can be significantly inspired by developmental biology models, we study regulatory patterns of tissue differentiation in normal and cancer conditions. Lastly, as the cardiovascular system is a paradigmatic example of an adaptive complex system, we apply our pattern recognition algorithms to explore normal/pathological conditions in cardiovascular patients.
Finally, our mission in translation and network medicine is to adopt advanced machine learning and network science approaches to integrate molecular networks and genomic profiles for the definition of personalised therapeutical plans and individualised drug treatments.
Future Projects and Goals
The goal of our group is to build novel advanced machine learning and graph/network science approaches for systemic analysis of biological systems like brain connectomes and, in general, molecular and cellular networks. We are particularly interested in network in developmental biology, cancer and neurodegeneration. In addition, we have particular interest in analysing the role of circadian and seasonal processes in biological systems and clinical medical variables.
Finally, our mission in translation and network medicine is to adopt advanced machine learning and network science approaches: 1) to identify hidden nonlinear patterns and combinatorial biomarkers in the gene/protein/lipid space; 2) to integrate molecular networks and genomic profiles for the definition of personalised therapeutical plans and individualised drug treatments.
Methodological and Technical Expertise
- Nonlinear adaptive approaches for unsupervised pattern recognition in multidimensional datasets based on the Minimum Curvilinearity theory.
- Topological Machine learning
- Methods for embedding and analysis of networks
- Methods for topological analysis of network wiring processes
- Methods for link prediction in complex networks
TF Consortium (including Carlo Vittorio Cannistraci).
A promoter-level mammalian expression atlas
Nature 507 (7493) (2014)
CV Cannistraci, G Alanis-Lobato, T Ravasi
From link-prediction in brain connectomes and protein interactomes to the local-community-paradigm in complex networks
Scientific reports 3 (2013)
CV Cannistraci, T Ravasi, FM Montevecchi, T Ideker, M Alessio
Nonlinear dimension reduction and clustering by Minimum Curvilinearity unfold neuropathic pain and tissue embryological classes
Bioinformatics 26 (18), i531-i539 (2010)
E Ammirati*, CV Cannistraci*, NA Cristell, V Vecchio, AG Palini, P Tornvall, …
Identification and Predictive Value of Interleukin-6+ Interleukin-10+ and Interleukin-6− Interleukin-10+ Cytokine Patterns in ST-Elevation Acute Myocardial Infarction
Circulation Research 111 (10), 1336-1348 (2012)
T Ravasi*, H Suzuki*, CV Cannistraci*, et al
An atlas of combinatorial transcriptional regulation in mouse and man
Cell 140 (5), 744-752 (2010)