Eldan Goldenberg's lab notebook

Learning how to use some new tools

Most of what I have accomplished in the past two weeks amounts to learning how to use the BN/PBN toolbox, and re-learning how to use Matlab. In a sense it's a bit frustrating—all this time not actually progressing in research—but it's also clearly necessary to do this sometimes.

The experiment replication I had almost finished with yesterday is now up and running, and replicates the published results very neatly. It turns out that my error wasn't in the code itself, but a simple typo in the input matrix.... I've also discovered that there is an existing function in the toolbox to do essentially the same thing as my code, which makes me feel a tad foolish, but really the point of the exercise was to learn the tools, so it wasn't entirely wasted effort. I was going to try and do the same thing in C++, but I feel like now that I'm over the hump of working out how to use it, the BN/PBN toolbox saves so much wheel-reinvention that it's just foolish not to use it, so unless I can find a C/C++ library to do the same sort of thing I'll stick to what I know.

This means I'm ready to start doing some 'real' work next week, and I'll sketch out one idea for this work behind the cut.

Li et. al. showed that the genetic regulatory network (GRN) observed in living yeast is impressively robust to perturbations. They did this by systematically generating singly-mutated versions of the real network (in each of which 1 link between nodes was either added, removed, or reversed between activation and inhibition) and measuring how much this changed the basins of attraction for the network. Most networks thus created showed the same overall behaviour as the real one, in that the vast majority of possible states lead to the same stable state, which corresponds to the state of gene expression in a growing yeast cell.

They also generated a batch of random networks that were similar to the biological one (in that they had the same numbers of nodes and links in each color [green for activation, red for inhibition] as the cell-cycle network), and found broadly similar patterns of robustness, even though the randomly generated networks tended to have a more attractors with smaller basins of attraction. This, to me, leads in to the general question of where the robustness comes from. There are several special properties of the whole set of random networks that Li et. al. used:

1. Number of nodes (all had 11)
2. Density of connections (3.1 per node)
3. Balance of inhibition to activation
4. Overall network type: they only created networks in which a node's activation is in effect determined by a vote among its incoming connections, which represent only a subset of possible boolean network architectures, and I know of at least one paper (Albert & Othmer, 2003) that uses a different architecture (in which any inhibition suppresses all activation) to model a GRN from biology.

I think this leads naturally to a next step of relaxing these constraints, individually or in groups, and probing the patterns of robustness that result.