Research Center for Computational Learning

Italy

Research Center for Computational Learning

Italy
SEARCH FILTERS
Time filter
Source Type

Castagnino N.,University of Genoa | Castagnino N.,Italian National Cancer Institute | Castagnino N.,Research Center for Computational Learning | Tortolina L.,University of Genoa | And 14 more authors.
Current Cancer Drug Targets | Year: 2010

The pathways downstream of ErbB-family proteins are very important in BC, especially when considering treatment with onco-protein inhibitors. We studied and implemented dynamic simulations of four downstream pathways and described the fragment of the signaling network we evaluated as a Molecular Interaction Map. Our simulations, enacted using Ordinary Differential Equations, involved 242 modified species and complexes, 279 reversible reactions and 110 catalytic reactions. Mutations within a single pathway tended to be mutually exclusive; only inhibitors acting at, or downstream (not upstream), of a given mutation were active. A double alteration along two distinct pathways required the inhibition of both pathways. We started an analysis of sensitivity/robustness of our network, and we systematically introduced several individual fluctuations of total concentrations of independent molecular species. Only very few cases showed significant sensitivity. We transduced the ErbB2 over-expressing BC line, BT474, with the HRAS (V12) mutant, then treated it with ErbB-family and phosphorylated MEK (MEKPP) inhibitors, Lapatinib and U0126, respectively. Experimental and simulation results were highly concordant, showing statistical significance for both pathways and for two respective endpoints, i.e. phosphorylated active forms of ERK and Akt, p one tailed =.0072 and =.0022, respectively. Working with a complex 39 basic species signaling network region, this technology facilitates both comprehension and effective, efficient and accurate modeling and data interpretation. Dynamic network simulations we performed proved to be both practical and valuable for a posteriori comprehension of biological networks and signaling, thereby greatly facilitating handling, and thus complete exploitation, of biological data. © 2010 Bentham Science Publishers Ltd.


Tortolina L.,University of Genoa | Tortolina L.,Italian National Cancer Institute | Tortolina L.,Research Center for Computational Learning | Castagnino N.,University of Genoa | And 13 more authors.
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) | Year: 2011

The signaling-network immediately downstream of the ErbB-family is very important in BC and other cancers, especially considering treatment of the excess of function of dominant onco-proteins with onco-protein inhibitors. We studied and implemented dynamic simulations of four downstream pathways. The fragment of the signaling-network we evaluated was described as a Molecular Interaction Map. Our simulations involved 242 modified species and complexes, 279 reversible reactions, 110 catalytic activities. We used Ordinary Differential Equations for our simulations. We started an analysis of sensitivity / robustness of our network, and we systematically introduced fluctuations of total concentrations of independent molecular species. We adopted mostly the strategy of a random sampling of 1000 cases for each instance of increasing numbers of perturbations. Only a small minority of cases showed an important sensitivity, the number of sensitive cases increased moderately for increasing numbers of perturbations. In most cases the effect of introducing virtual mutations and virtual onco-protein inhibitors was more important than the effect of randomly introduced perturbations, this suggests an acceptable robustness of our network. The importance of our work is primarily related to the fact that the complexity of the 39 basic species signaling-network region we analyzed is of difficult intuitive understanding for a "naked" human mind. Dynamic network simulations appear to be an useful support for an "a posteriori" mental comprehension by a cancer researcher of the behavior of a network of this degree of complexity. The present report suggests the feasibility of a computational approach even in the presence of a multiple number of uncertainties about parameter values. © 2011 Springer-Verlag Berlin Heidelberg.


Tortolina L.,University of Genoa | Tortolina L.,Italian National Cancer Institute | Tortolina L.,Research Center for Computational Learning | Castagnino N.,University of Genoa | And 11 more authors.
Current Cancer Drug Targets | Year: 2012

This review article is part of a special Current Cancer Drug Targets issue devoted to colorectal cancer and molecularly targeted treatments. In our paper we made an attempt to connect more basic aspects with preclinical, pharmacological / therapeutic and clinical aspects. Reconstruction of a Molecular Interaction Map (MIM) comprising an important part of the G0 - G1 - S cell cycle transition, was a major component of our review. Such a MIM serves also as a convenient / organized database of a large set of important molecular events. The frequency of mutated / altered signaling-proteins indicates the importance of this signaling-network region. We have considered problems at different scale levels. Our MIM works at a biochemical-interaction level. We have also touched the multi-cellular dynamics of normal and aberrant colon crypts. Until recently, dynamic simulations at a biochemical or multi-cellular scale level were considered as a sort of esoteric approach. We tried to convince the reader, also on the basis of a rapidly growing literature, mostly published in high quality journals, that suspicion towards simulations should dissipate, as the limitations and advantages of their application are better appreciated, opening the door to their permanent adoption in everyday research. What is really required is a more interdisciplinary mentality and an interdisciplinary approach. The prize is a level of understanding going beyond mere intuition. © 2012 Bentham Science Publishers.

Loading Research Center for Computational Learning collaborators
Loading Research Center for Computational Learning collaborators