Saturday, April 08, 2006

Systems Biology

This month's Nature Reviews Molecular Cell Biology is dedicated to "Systems" Biology. This is a redundant term but the distinction is necessary today since most of biology operates under a reductionist paradigm rather than focusing on systems. It's time to start doing so because the key to understanding and manipulating the phenotypes of cells, organisms and populations (the only worthwhile reason for studying for biology in the first place) is essentially an exercise in modeling complex systems. So read it. Also, http://www.systems-biology.org/ is a great resource, with links to hot new papers, conferences etc. The First Conference on Systems Biology of Mammalian Cells...interesting.


6 comments:

Anonymous said...

To paraphrase:"The only worthwhile reason for studying something is so you can manipulate it."??
I don't agree that the development of technology should be the goal of all of biology. For example: What about the understanding that as a species we have more biological similarities than differences? I can think of lots of biological scientific knowledge that is more important to the world than the latest way to manipulate it, make a drug ect. Changing the way we look at ourselves and biology shapes our understanding and outlook and allows us to make more rational decisions.

Bayman said...

I agree. That's why I said manipulating AND UNDERSTANDING. So understanding I would agree is probably most important. However the majority of cellular/molecular biology research funding in North America and most of the rest of the Western world is directed at developing new treatments for cellular disease. This means manipulating the cellular phenotype (ie quiescent vs proliferating vs dead vs alive invasive vs conductive, etc, etc.....in a cell-speicific manner. So most employed researchers in the field work on this in at least some capcity. Of course it is much more important to understand how biological systems work on all levels, to understand humanity etc. Theoreticaly biological systems show self-similarity and so determining the rules that govern cellular networks would be applicable to the behavior of human populations, neural networks, ecosystems, or bacteria on Mars. Of course this works vice versa, but its harder to do controlled experiments on ecosystems for example, vs cells.

Bayman said...

Another point now that I think about it. If we're looking to molecular biology for moral guidance we've got problems. It's pretty obvious if you want to see it that humans have more similarities than differences, no amount of comparing signal transduction pathways is going to get anyone any closer to the truth or make people love each other. When was the last time molecular biology helped you make a more rational decision? Will there ever come a day where we can seek guidance by asking: "What would p53 do...?" The only hope in making science useful beyond a tool for technology development is to get away from a reductionist viewpoint and build more holisitic models of understading.

Anonymous said...

Are you suggesting that a yeast interactome is not reductionist?
Reducing yeast to a network of protein protein interactions without any thought to the subtleties of these interactions is just as reductionist as looking at any one particular interaction in detail. What about metabolites and lipids? What about RNA?
Without reductionism we would be left with looking at yeast populations in their natural habitat. pretty unscientific. especially since their natural habitat is yo' momma.
I don't make decisions based on p53 everyday but knowing that cancer has not only a genetic basis but that this basis is often the same makes a huge difference to the way humanity looks at the disease. It's easy to say duh of course it is genetic and has common causes. But that is only from research into things like p53 that we know this. Just knowing that cancer can be classified ect takes the 'magic' out of it and changes your reaction if you were diagnosed into a rational perspective. Even if treatment was never improved based on this knowledge.

Anonymous Coward said...

I think systems biology is the way to go. Combining robotics, high throughput, and autonomous computer analysis is got to be more efficient then redundent, biased human inquiries. Can computer have scientific insight? clearly no, but insight is a human shortcut necessary because we have limited means. I can't study 40 000 genes with trillions of interactions at once. What will limit systems biology is computability. This is non trivial and I recommend the stuff Penrose wrote on computability to fully comprehend this. But basically other than the fact that we still don't have anything powerfull enough to simulate a full living cell with all the interactions of lipids, sugars, proteins, environment etc, even if we had this all powerfull computer, it may still not be computable. There are such things as problems that algorithms cannot solve (N polynomials). There are quantum effects within a cell that are non deterministic, hence one cannot ever have an initial state and deduce what will happen at time t. So we may never be able to simulate life.

Bayman said...

Good points by anonymous, especially about yeast interactomes...I agree this is overly reductionist and mapping yeast interactomes has gotten way too popular too easy beacuse people can just throw out some crazy network diagrams and talk like they're modeling a cell. The most valuable results from these kinds of studies will probably be the stimulating progress in the field of bioinformatics. But in terms of real biological systems, we all know it's not that simple, proteins for example, exert influences on the cell through enzymatic activities (of which each protein may possess more than one), physical properties such as complex scaffolding, cytoskeletal or allosteric effects, and their spatio-temporal distributions. So the real huge challenge is understanding network models that take all of these factors into account...not easy because these networks will be massively interconnected and multi-dimensional. But doable.

Anyway I think this kind of understanding would have some even very basic implications for molecular biologists. For example it's the only way to really understand a lot of the experimental work we do. What happens when you knock down a protein in a cell? Overexpress one? How about knock it out in a mouse? Inhibit one or more enzymatic activites with a small molecule? Right now it's all trial and error, and we just keep trying individual genes until we get a phenotypic change that interests us, and then we make up some story to tell all our scientist buddies about how this tells us about the important role that this particular gene plays in this or that biological process. Of course most of it is probably incorrect, or at least viewed from the wrong point of view. Anyway point is if you understood cellular networks you could talk in a much more intelligent way about what the actually effects are of knocking down a protein for example. By analogy to the domino effects of extinction of a single species on an entire ecosystem, removing a gene that interacts with a cellular network in many different ways will effect that network's output (ie phenotype) in ways you could never predict without using a comprehensive paradigm.