Home
Mathematical models can serve as a tool to formalize biological knowledge from diverse sources, to investigate biological questions in a formal way, to test experimental hypotheses before performing them in a wet lab, to predict the effect of perturbations applied to a biological system and identify its different modes of functioning.
Using a model of initiation of the metastatic process, we present a pipeline of computational tools that performs a series of analyses in order to increase its predictive power. As many entities in cellular biology are represented by networks of interactions, we start by analysing the structure of such a network. Some sets of nodes that participate in a particular process can be identified and proposed as gene targets for combined therapies, while trying to minimize possible side effects. Next, we explain how to translate this network into a mathematical object, specifically a logical model, and how some robustness analyses can be applied to show that the model, in most cases, is only sensitive to a small subset of perturbations. We argue that the logical model can be viewed as a tool to identify the main players and understand better the biology behind some cell fate decisions. One way to explore these decisions is to reduce the network to its building blocks to extract the biological motifs that control the cell fate decision process. We present two approaches: one by reducing the number of variables and another one by grouping nodes into modules. Another analysis concerns the solutions of the wild type model: based on the list of all stable states, the variables of the model can be classified per their contributing grade to a given output. We explain how to compute the probabilities to reach a given cell response, how to identify probabilities in genetic interactions and how to conclude on synergistics effects of those, let it be synthetic lethality or enhancement of two alterations. Finally, we show how the solutions of a mathematical model can also be compared or even trained on experimental data, exploring ways to match data to model solutions. Similarly, data can be mapped on the network using a standard approach (mapping the expression of an entity on the network) or using a modular approach (mapping the expression of the targets related to the gene).
All these approaches aim to verify that the model complies with what is known about the processes it describes and to increase the predictive power of these models. These approaches are implemented in easy-to-use tools, including BiNoM, OCSANA, GINsim, MaBoSS, Lemon-Tree and VidaExpert.
Full tutorial can be followed on the dedicated Tutorial webpage
Description of the main results of our pipeline can be found at:
Arnau Montagud, Pauline Traynard, Loredana Martignetti, Eric Bonnet, Emmanuel Barillot, Andrei Zinovyev, Laurence Calzone; Conceptual and computational framework for logical modelling of biological networks deregulated in diseases, Briefings in Bioinformatics, bbx163, https://doi.org/10.1093/bib/bbx163
This repository can be refered to with its own DOI:
