Systems and Computational Biology

High-throughput omics data and computational approaches are today providing a key to disentangle the complex architecture of living systems. The integration and analysis of data of different nature allows to extract meaningful representations of signaling pathways and protein interactions networks, helpful in achieving an increased understanding of such intricate biochemical processes.

Mathematical modeling and simulations

Mathematical biology aims at understanding biological phenomena that prove too complex to an analysis without the use of the exact sciences. With mathematical modeling (or computational) we set a quantitative description of the observations obtained by experiments in "traditional" laboratories. Adherence to the experimental data is the necessary condition for the mathematical model while the predictive power of the latter is its added value. The primary objective of this activity is to get results that have a predictive value with impact directly expendable either in the understanding of basic biological principles or as practical advices for the medical clinic. A second objective, methodological, is to provide a set of mathematical tools to describe and study the dynamics of complex biological systems to more levels of detail (i.e., intracellular, extracellular, systemic). A final important objective is to establish a point of reference for all CNR departments of the Department of Life Sciences and promote an interdisciplinary exchange between biology and mathematics.

Simulation of the immune system dynamics

Mathematical modeling and computer simulations have found application in nearly all scientific disciplines. Biology, today, is no exception. High throughput data-acquisition methods in molecular and cell biology are constantly driving the field of bioinformatics to develop new tools to help in understanding biological complexity at the molecular scale. At the cellular level, cells themselves can be considered as units or actors. The goal of the application of most mathematical and computational models applied to the study of infectious diseases is to seek to unravel the “laws” of nature related to diseases. In the clinical setting, a vast array of analytical and diagnostic tools is presently available to capture data and to construct well-tailored mathematical or computational models. Therefore, time is ripe to exploit the predictive power of mathematics in vaccinology.

Immunology is prime example of a medicine-related field that stands to gain much from applied mathematics and modeling. The immune system is recognized to be one of the most complex systems that science strives to understand, and new approaches such as the science of complexity can be usefully applied.

One of the main activity of the DBU focuses on a computer simulation tool constructed using the basic principles of immunology as from antecedent conditions; they arise through the interaction of a large number of cooperating components (i.e., the paradigm is those of the Agent-Based Modeling). Simple rules can be repeatedly applied to a large number of simple elements, so that in the end, aggregate or macroscopic variables can be observed not just as the sum of the elementary components, but instead as pure unpredictable emergent phenomena.

Modeling the lymphatic system network

Development of a three dimensional model of the lymphatic system. The aim is to produce a mesh that allows us to study and reproduce cancer cells migration. The cells travel through lymphonodes and vessels. In addition to this, we want to represent the whole process as it connects with the circulatory system.

Data integration and network biology

Complex biological systems may be represented and mathematically analyzed as computable networks, using graph theory. DBU is involved in computational methods for the reconstruction and analysis of protein-protein interaction (PPI) networks, gene regulatory networks (GRNs, DNA-DNA and DNA-protein interaction networks), signaling networks (signals transduced within cells or in between cells to form complex signaling networks), metabolic networks and boolean networks under both synchronous and asynchronous updates.

Statistical mathematics and machine learning for computational biology

Estimation methodologies for high dimensional models and machine learning. These methodologies are successfully applied in computational biology. In particular, we work on creating surrogate statistical models and emulators whose computational cost is negligible compared to fully-detailed mathematical models. Given the complexity of the models considered in biology, We make full use of statistical and machine learning algorithms, from ensemble methods such as Random Forest to Gaussian processes just to name few, in order to find the most suitable statistical model in terms of computational cost efficiency and predictive power.

Bioinformatics

Digital Biology Unit

Research