### A small celebration

A much-delayed post here, because I've been sweating over a conference paper for the last couple of weeks. Deadline yesterday, and we met it! Albeit with my co-author in Australia staying up until 2am to finish the stats, and becoming decidedly monosylaabic in the process. But the main thing is that it's done and submitted and out of my hands.

In the paper we look at the relationship between network motifs and network dynamics. The seminal paper in this area was published in 2002 [1], but the idea has really taken off. Basically the idea is that a large complex network can be broken down into small network motifs consisting of three or four nodes, which carry out a very specific task - being a logical 'and' gate or providing negative feedback control, or whatever. The hope is that if we can characterize enough of these motifs we will be able to get a handle of the way large networks behave by looking at the way their motifs behave.

This is a lovely idea, and offers the first real hope of being able to understand the way in which things like genetic regulatory networks operate other then simply prodding them and seeinghow they respond. Statistically over-represented motifs have been found in lots of different kinds of networks, biological and otherwise, and the fact that the same motfs can be found in very different networks means, according to some researchers, that they are evolutionarily conserved.

However, when you're looking at groups of three or four nodes there just aren't that many combinations in which they can occur. In addition, looking at huge numbers of nodes and the links between them means that you have to interpret your statistics with caution; if the chance of a particular combination occurring is 1 in a thousand (0.001, usually considered statistically significant) you only need to make 500 observations to have a good chance of seeing that combination. Ans with large nets we're looking at lots of combinations! So while I think the network motif idea is a very neat one, and I'd like it to be the basis for useful analysis, I think it remains unproven.

In out study we took a reverse approach and generated computational networks with and without interesting dynamic behaviour and then looked to see what the structural differences between them were. There was certainly no increase in numbers of motifs, although we did see a few. What did change dramatically was the numberof feedback loops in the networks. Since feedback loops are known to be important in the generation of network dynamics, this makes sense, and we're continuing to follow it up.

We concluded that you can have interesting dynamic behaviour without the presence of network motifs. This doesn't mean that real networks don't have them, but it does mean that they don't have to have them, and hints that the relationship between network topology and dynamics is not going to be easily untangled. But we're working on it! Fingers crossed that the paper gets in...

So, this was a bit of a diversion from aging, but I'll get back into it next week, when my brain is less full of triads and simple cycles.

[1] Milo, R., Shen-Orr, S., Itzkovitz, S., Kashtan, N., Chklovskii, D. & Alon, U. (2002). Network motifs: Simple building blocks of complex networks. Science 298: 824 - 827.

In the paper we look at the relationship between network motifs and network dynamics. The seminal paper in this area was published in 2002 [1], but the idea has really taken off. Basically the idea is that a large complex network can be broken down into small network motifs consisting of three or four nodes, which carry out a very specific task - being a logical 'and' gate or providing negative feedback control, or whatever. The hope is that if we can characterize enough of these motifs we will be able to get a handle of the way large networks behave by looking at the way their motifs behave.

This is a lovely idea, and offers the first real hope of being able to understand the way in which things like genetic regulatory networks operate other then simply prodding them and seeinghow they respond. Statistically over-represented motifs have been found in lots of different kinds of networks, biological and otherwise, and the fact that the same motfs can be found in very different networks means, according to some researchers, that they are evolutionarily conserved.

However, when you're looking at groups of three or four nodes there just aren't that many combinations in which they can occur. In addition, looking at huge numbers of nodes and the links between them means that you have to interpret your statistics with caution; if the chance of a particular combination occurring is 1 in a thousand (0.001, usually considered statistically significant) you only need to make 500 observations to have a good chance of seeing that combination. Ans with large nets we're looking at lots of combinations! So while I think the network motif idea is a very neat one, and I'd like it to be the basis for useful analysis, I think it remains unproven.

In out study we took a reverse approach and generated computational networks with and without interesting dynamic behaviour and then looked to see what the structural differences between them were. There was certainly no increase in numbers of motifs, although we did see a few. What did change dramatically was the numberof feedback loops in the networks. Since feedback loops are known to be important in the generation of network dynamics, this makes sense, and we're continuing to follow it up.

We concluded that you can have interesting dynamic behaviour without the presence of network motifs. This doesn't mean that real networks don't have them, but it does mean that they don't have to have them, and hints that the relationship between network topology and dynamics is not going to be easily untangled. But we're working on it! Fingers crossed that the paper gets in...

So, this was a bit of a diversion from aging, but I'll get back into it next week, when my brain is less full of triads and simple cycles.

[1] Milo, R., Shen-Orr, S., Itzkovitz, S., Kashtan, N., Chklovskii, D. & Alon, U. (2002). Network motifs: Simple building blocks of complex networks. Science 298: 824 - 827.