Wednesday, May 16, 2007

Systems Bio. GAH.

An idea:

Biological networks are, by and large, scale-free networks (with a few high connectivity nodes and many low connectivity nodes, rather than mostly average-connectivity nodes). These seem to evolve continuously and quickly, rather than in a punctuated way (which is how random networks evolve, btw). Which, at first glance, goes against a "punctuated equilibrium" model until you realize that in punctuated equilibrium you aren't just modelling the evolvability towards one trait but rather the sequential evolution of many traits. ALSO, many protein people (and DNA people?) note significant homology between varying domains of different proteins and propose a mix-and-match model of evolution. FINALLY, transcription factors are overrepresented in genes which have significantly diverged from the human/chimp, human/ape or human/primate common ancestor. A couple ideas:

1) Do scale-free networks give you "punctuated equilibrium" when faced with a changing environment/target? If this is true, it would navigate the apparent difference above and explain why we observe scale-free networks with punctuated equilibrium evolution in nature.

2) What happens to network evolution when you preferentially change "transcription factors" (nodes which effect the state of other nodes)? This might shed some light on specifically human evolution and what makes us different from our closest evolutionary cousins.

3) Do random networks evolve into scale-free networks when you can add connections biased towards "transcription factors"? If this is true, it would go some distance to explain why biological networks are by and large scale-free.

EDIT:

So as I have one of the most Biology-heavy backgrounds in the Systems Bio class and certainly in the PCBio program (by which I mean I am either the first or second most Biology-heavy person in PCBio, depending on how you rank Pearl and myself), it seems odd at first that I want to think about computer science. But then again, I was caught explaining BOTH of the computer-science papers to my cohort, which puzzled me at the time.

I stopped being puzzled when I started going on a rant about STELLA and how horribly, awfully boring and ridiculous a programming language it was (but how it was particularly designed to allow one to easily create a dynamic network). I think I started talking about linked lists after that.

Which reminded me of Mr. Hiedler's class. And cults.

Oh Montgomery Blair.

No comments: