The student will start from existing biological network data and network models provided by other partners (F. Zeldenrust, P. Poirazi, G. Debregeas) and investigate the specificity of the relationship between local/global topology and network activity/computation. For this, we will thoroughly evaluate and combine existing quantitative measures of graph topology (e.g., node degrees, distance-based connectivity, motifs, hierarchy-based measures like Rent’s exponent) and graph similarity (e.g., graph kernels used in machine learning) and devise new ones that are more appropriate to biological networks. He/she will use these measures in the provided network simulation models to quantify the sensitivity of network activity and information processing to variations in these measures, or in other words, how well ‘abnormal’ behaviour can be explained based on variations in these structural network measures. From these experiments, we will derive probabilistic models that express the relationship between local/global parameter variability and variability in the network activity patterns and information processing.
More information: https://www.ugent.be/en/work/vacancies/scientific/phd-student-lo8js
1. Determine the relationship between topology and network activity patterns.
2. Determine the relationship between topology and network information processing.
3. Identify useful quantitative measures and probabilistic models that express this relationship and use them to identify structures that will display ‘abnormal’ behaviour.
1. Radboud University, Fleur Zeldenrust: M11-13: get acquainted with network data and network models.
2. IMBB, FORTH, Panayiota Poirazi, M27-30: apply developed models on pre- and post-learning networks.
Enrolment in Doctoral degree(s): You will be enrolled at Ghent University PhD programme