Talk:Systems biology/Archive 1

Starting a talk page on systems biology
Wikipoedists, I am starting a talk page on systems biology. I work at a place called the Institute for Systems Biology, and have worked with Lee Hood for the last 8 years. I think at we are one of the Institutes who are defining what it is to do Systems Biology. I also think that the definition is still in flux as we see what we can do and what we can't. One reason to start this discussion here is that I have not heard or read some of the terms used in this article and don't agree with other points made in the article. I think it would be a good idea to have a discussion of what the article should say.

First of all, systeomics. I have never heard of systeomics. Not that that should mean anything, but you know, even though a friend of mine, Eugene Kolker is the editor in chief of the journal Omics, I think I can speak for a lot of people in the field, in saying that we are tired of the proliferation of all these omics. Enough. Lets come up with a different term, or, at least, can whoever said syteomics is a synonym for Systems Biology reference it? Was it in a Kitano article? What journal was it used in?

Then, the article says that systems biology hopes to understand how the whole works by studying all the parts. In fact, this is how reductionist biology has been done for the last 50 years, and is not really how we are trying to use systems biology.

One of the major shortcomings of reductionist approaches is that Nature is too complex to put the whole back together from all the individual parts. Nature is too redundant. There are too many routes to follow to rebuild the whole from the parts.

I came to this conclusion when I found that, at a molecular level, the only thing different in the way two cell types responded to vitamin D was that Jun D was regulated differently (one of the refs would be JBC in 1994 or 1995). The Jun family of proteins are transcription factors that regulate the expression of hundreds of proteins, many of which are transcription factors themselves.The whole thing is too combinatorial to try to put it all back together from the parts. You need to do it differently.

I think what systems biology does, and this is why systems biology was not really doable in the past, is use high throughput techniques to look at how everything changes in response to a perturbation. We couldn't do it before because we didn't have the parts list, the list of all the potential genes that can be encoded by the genome. This was accomplished by doing the human genome project, which was a dsicovery science project, not a systems biology project.

Discovery science is where you define all the parts of a system without have a specific question about how they work. Most of us knew that at some point we would need the whole genome sequenced, in fact, we would need a bunch of genomes sequenced, so why not just do it all now (those of us in Academia owe Craig Ventor a word of thanks for giving us a little nudge to get it done). That was why many researchers were opposed to doing the human genome project; they felt that the sequences that we needed to know would be discovered by sequencing the regions determined to be important in answering specific hypotheses. Discovery science gives us a list of all the genes in the genome, all the mRNAs in the transcriptome, all the proteins in the proteome, and all the interactions in the interactome (we don't have a great high throughput way to look at the interactome yet, but two hybrid studies are a start). Discovery science projects give you the potential parts of a system.

We also didn't have the computational power that we needed (see moore's law and take for example of the 300,000 plus processors we (United Devices, IBM and the ISB) have folding all the sequenced orfs) and not many biologists were actively working with engineers and physicists who are used to looking at massive amounts of data. In fact, a lot of us went into biology because we didn't particularly like math. I think that the days of the math phobic biologist are drawing to a close.

In systems biology, you, this cross-disciplinary scientific unit, use this complete parts list to make a model of how a system works. Several investigators (actually many investigators), including scientists at the ISB, are developing software tools that allow us to describe these interactions and networks mathematically, and to tweeze the rare important interaction out of the whole hair-ball of connections. This can be approached in a number of ways, for instance using ordinary differential equations or statstical mathematics, or even quantum nechanics (Hamid Bolouri edited an interesting book describing these methods and the advantages and disadvantages or each, I don't have the ref at my fingertips, but I'm sure a search for sys biol and bolouri at Amazon will bring it up).

By carefully modeling the system we can predict what perturbations will do, in other words, we can make a hypothesis, and then go back to the bench, and test the models. As we test and change and refine our models, we can come closer to understanding how systems work. By observing all of the things that change with each perturbation, properties of the system shouild emerge which would not be apparent otherwise.

Next in the article, I don't think that interactomics is the same as systems biology just like I don't think systeomics is systems biology. It is one of those omics that you use in systems biology.

Next, in the from genes to patients section, the explanation seems backwards to me. What we hope is that systems biology will help us identify unique biomarker patterns that can tell us what, exactly, caused your cancer and give us a clue on how to address your personal cancer, rather than just killing everything that is growing.

Next, I don't think that the discussion of hypothesis-driven and data-driven science is correct when trying to describe systems biology.

Maybe its just semantics, but we believe that systems biology is hypothesis driven, and hypothesis driven science does not try to decompose all the parts to understand what is going on. I think that the definitions of hypothesis driven science and data driven science given here both describe reductionist approaches to science as opposed to systems biology. This is the hard part and the major departure from the way we thought about and did science in the past; We have to let go of all the parts and start looking at and quantifying how the whole is manifested. Its the connections between the parts that is important.

As I said, systems biology is hypothesis driven and the hypothesis is a testable model. The model is going to be molecular, and should be described mathematically (something like biospice or e-cell will eventually work). Iterative testing and remodeling take place until what we predict is the same as what we see.

But decomposing the system into separate subsystems takes us back to where reductionism fails. All of these parts, (which will ALWAYS be molecular in systems biology), not just in an extreme case, are connected to each other and its is the connections between the parts that are the emergent properties of the system that we must find to understand the systems biology of a cell or other biological unit.

So... that is an attempt to start a discussion on what systems biology is. We have been discussing this in Lee's labs for the last 8 years. Over that time the definition has changed and we still don't have the final definition of what systems biology is. I hope the discussion on wikipedia can also further our understanding of systems bioloogy. srlasky 06:13, 2005 Mar 29 (UTC)

Well, I attempted to start a discussion on what systems biology is, since its my opinion that what has been described in this article so far is not systems biology. I will take the lack of response to mean that people won't mind if I start putting in the current definition of systems biology as developed at the Institute for Systems Biology. It will probably take a few weeks to make the changes because they are extensive. srlasky 03:32, 2005 Apr 25 (UTC)

The current definition and description of systems biology seems to be focused on the mathematical world. Is it always true that it is necessary to mathematically model a system to understand it? Causal Modeling has been quite effectively used to study the associations between biological cause and effect relationships, and can be quite effectively used in areas where there isn't sufficient kinetic information to build a mathematical model. It would seem appropriate to treat causal modeling of biological systems alongside mathematical modeling.