I have been invited to give a talk at Yamanashi University on the contribution of Computer Science to biological modeling (especially when it comes to gene regulatory networks). It was both challenging to introduce my research to a brand new audience and exciting to discover a (beautiful) place like Yamanashi.

This talk was part of my activity as JSPS invited researcher at Inoue Lab, National Institute of Informatics (Tokyo).

Invited talk at Yamanashi University

Invited talk at Yamanashi University

Hidetomo Nabeshima was chairing my Invited talk at Yamanashi University. Thanks to him!

Yamanashi University

One of the announcement of my talk at Yamanashi University

View from the 8th floor of some building at Yamanashi University

Yamanashi countryside

Sengataki Falls in the area of Yamanshi

Details

• Place: Kofu, Yamanashi University
• Date: 2014/06/18
• Duration: 1 hour and 30 minutes
• Attendance: 30+ persons (researchers, PhD students)

Abstract

Analyzing the Dynamics of Large Biological Regulatory Networks with Process Hitting
(joint work with Maxime Folschette, Loïc Paulevé and Olivier Roux)

Regulation is a key aspect of biological systems, all the way from the molecular scale to the ecological scale. In order to make relevant analyzes of such systems, it is then crucial to get a precise understanding of this phenomenon. This is precisely one of the main goals of systems biology.

The modeling of biological regulation can be divided into two main trends. The first is based on ordinary differential equations involving the quantitative expression of the interacting components. However, as data in a biological context are often more qualitative than quantitative, it is meaningful that another trend, based on qualitative modeling, emerged in the late 1960s. The principle of this modeling framework, introduced as synchronous Boolean networks by Kauffman and asynchronous Thomas’ networks, is to represent genes as Boolean variables. These Boolean paradigms may appear to be simplified models, but they led to significant results about the behavior of networks, such as cyclicity or steady states. Moreover, these models have been extended over the years, for example, to consider additional levels of expressions.

Nevertheless, classical analysis approaches generally rely on the exploration of the state space and parameter identification requires some indirect reasoning. This becomes tricky when the model grows beyond 10 interacting components; because of the combinatorial explosion, it is extremely difficult to handle large, realistic regulatory networks.

In order to address this scalability issue, we recently introduced a new framework, called Process Hitting. Establishing relationships between the components at the most atomic level possible, the Process Hitting opens the way to many static analysis and abstract interpretation methods to study complex dynamic properties.

In this talk, we will present the Process Hitting modeling approach and the methods we designed to analyze its dynamics. We will illustrate its benefits on a case study and give some benchmarks.