In silico network reconstruction
It is day one of Just Science week and I want to tell you about a recent paper that was published in BMC Systems Biology by Rui Alves and Albert Sorribas. It is about a general approach to integrate information to come up with models for cellular pathways. What does this mean and why is this important ?
Increasingly the scientific knowledge is being stored in databases (literature, protein structures, gene expression, protein-protein interactions, protein-DNA interactions, etc). The general idea behind the work described is that we should be able to use the accumulated information about cellular pathways to extract models of how the cell's components interact to preform their functions. By models I mean a formal representation that can tell us how the components' concentrations and activities change with time.
There are several works already dealing with this problem of trying to reconstruct cellular networks from large data sources but I found this article particularly interesting because it uses so many of these methods.
To give you an idea I reproduce below figure 4 of the paper with a diagram of the method (click to zoom in):
The authors have pulled in experimentally known interactions and combined them with putative interactions obtained from docking and phylogenetic based predictions. These predicted networks are then converted to several possible mathematical models that are examined under different parameter conditions and compared with known experimental values.
This method should be particularly suited for a case when some of the genes in the pathway are known and there are experimental measured outputs for the pathway that can be compared with the predictions from the putative pathway models.
Ideally this whole procedure would be fully converted into an automatic pipeline that could be used by people that are not so familiar with the tools.
I will try to stick with the same theme during the week, hopefully covering different methods to achieve the same thing.