Network Inference

When we (implicitly or explicitly) reconstruct signaling networks in biology we generally do so by trying to understand the consequences of when we break something (e.g. gene deletions) or by mapping physical interactions (e.g. ChIP for gene regulatory networks). Both methods have their problems.. In the former it is difficult to entangle direct from secondary effects. The later does not work well for networks based on transient interactions such as phosphosignaling.

Network reconstruction from temporal -omics data is a promising alternative. Current methods are not adapted to data of high temporal resolution - which, however, likely are the best data in the case of phosphosignaling. 

My plan:

1. Construct candidate networks via a graph based approach (Steiner Trees)

2. Prioritize networks using Dynamic Bayesian Networks

3. Evaluate the global fit via ODEs

Here`s my attempt at step 1:

https://doi.org/10.1101/2023.04.26.538385 

https://github.com/m-i-c-h-ael/Kanshin_NetwInf