Directed and undirected network inference by message passing and applications to gene regulation

The European network "Statistical Physics Approaches to Networks Across Disciplines" (NETADIS) is looking to recruit excellent Early Stage Researchers (ESRs) for attractively funded three-year PhD research projects at 9 leading European partner institutions. This position at King's College London should be taken up on 1 September 2012 or as soon as possible thereafter. Applications arriving after 15 March will be considered where possible.

Candidates are eligible to apply if they have an excellent first degree that qualifies for PhD study in a relevant discipline (e.g. physics, mathematics, computer science, engineering) with preferably a strong background in statistical physics, and less than four years of research experience. At the time of recruitment they must not have resided or carried out their main activity in the U.K. for more than 12 months in the last 3 years. Any PhD course fees will be met from the training expenses contribution to NETADIS, subject to budget restrictions.

NETADIS aims to train a cadre of future research leaders in advanced methods of analysis, inference, control and optimization of network structure and dynamics, to maximize the impact of statistical physics approaches across a broad range of application areas. It is funded by the EU Marie Curie Actions. Each NETADIS project is defined by a principal research theme (inference; dynamics; optimization and control) and a principal application domain (systems biology and neuroscience; information technology; socio-economic systems and finance; network phenomena in physical (laser) systems).

This project deals with inference of gene regulation networks. Major challenges are (i) sparse-model learning, since most regulatory networks are sparse, with one variable being directly influenced only by a small set of other variables, (ii) combinatorial inference, since the regulation of one variable by others may include nontrivial combinatorial effects in combining the single regulators, (iii) the handling of missing variables, i.e. of components of the system which are not determined by the measured data. Given the large size of real biological networks, approximate algorithmic approaches are needed; exact algorithms for network reconstruction are restricted to very small problems. Recent advances in the statistical physics of disordered systems have led to first promising results. This project will address the outstanding challenge of exploiting these for the full benefit of applications, here specifically inference of signal-transduction networks in cancer cell lines from multiple perturbation data, and of residue-contact networks in proteins and protein complexes.