Dismiss Notice
Join Physics Forums Today!
The friendliest, high quality science and math community on the planet! Everyone who loves science is here!

Systems Biology

  1. Apr 26, 2008 #1
    "Systems biology" is a phrase that gets thrown around a lot, though it never seems to be completely clear what it actually means. Wikipedia defines it as:

    "Systems biology is a relatively new field that focuses on the systematic study of complex interactions in biological systems"

    What do you think this means? If you had been writing the wikipedia article, how would you have defined it?

    Use of the word "systems" in the life sciences seems to be pretty inconsistent. For example, when I hear "systems biology" I think of genomics. However, when I hear "systems neuroscience" I think of electrophysiology. The former studies things much smaller than cells and the latter studies networks of many cells. Though naively from hearing the names you would probably think that systems neuroscience was a subset of systems biology.
    Last edited: Apr 26, 2008
  2. jcsd
  3. Apr 26, 2008 #2


    User Avatar
    Staff Emeritus
    Science Advisor
    Gold Member

    I think it's only new to those who cut their teeth on cell or molecular biology and have forgotten that there's a world of biology beyond their cells in a dish. It's basically going back to the whole organism and remembering that systems in the body all interact. For example, if you want to study cardiovascular physiology, you can't do it in a vacuum of ignorance about endocrinology, neuroscience, renal physiology, etc., because those all impact on the system you're studying. And, there is also the other issue of addressing these interactions from the whole animal down to the cellular level.

    So, no, it's not genomics or electrophysiology...those are very specific fields. Systems neuroscience, for example, would be studying the neural circuitry involved in a particular behavior or function (the whole system of interactions), and you would use a variety of approaches to study it...some of it may include electrophysiology, molecular biology, anatomy (i.e., tract tracing), microscopy, pharmacology, etc. It's basically getting away from the "gene jockey" trend that had become so prevalent.
  4. Apr 26, 2008 #3
    That sounds really good as a definition. Though I don't think it really follows the way I hear the words used. All my systems biology friends (of whom, I have none) talk about genomics/proteomics.

    In systems neuroscience (my field) people do use lots of different methods as you say, but the primary one is always electrophysiology. Can you name a single systems neuroscientist who doesn't do some amount of electrophysiology?

    At least in neuroscience this may change in the future as various optical techniques get more powerful and widespread. Though for now, electrophysiology seems to be the method of choice for nearly everyone.
  5. Apr 28, 2008 #4

    Andy Resnick

    User Avatar
    Science Advisor
    Education Advisor

    "systems biology", as I have seen it used, is taken to mean analysis of networks- signalling networks within a cell, the proteome, the metabolome, the genome, etc. etc. I have hardly ever seen anyone use the term 'systems biology' in context with an organ, organism, or ecosystem (which is different from saying the term does not apply; obviously it does).
  6. Apr 28, 2008 #5
    Here's an example of a systems approach in developmental biology. Eric Davidson's group is studying the gene regulatory network needed to pattern the sea urchin embryo, but is also comparing that network with those of other creatures to learn how the network changes though evolution.

    Oliveri P, Tu Q, Davidson EH. Global regulatory logic for specification of an embryonic cell lineage. Proc Natl Acad Sci U S A. 2008 Apr 14; [Epub ahead of print]

    Hinman VF, Davidson EH. Evolutionary plasticity of developmental gene regulatory network architecture. Proc Natl Acad Sci U S A. 2007 Dec 4;104(49):19404-9. Epub 2007 Nov 27.

    Smith J, Theodoris C, Davidson EH. A gene regulatory network subcircuit drives a dynamic pattern of gene expression. Science. 2007 Nov 2;318(5851):794-7.

    Livi CB, Davidson EH. Expression and function of blimp1/krox, an alternatively transcribed regulatory gene of the sea urchin endomesoderm network. Dev Biol. 2006 May 15;293(2):513-25

    Yuh CH, Dorman ER, Howard ML, Davidson EH. An otx cis-regulatory module: a key node in the sea urchin endomesoderm gene regulatory network. Dev Biol. 2004 May 15;269(2):536-51.

    Hinman VF, Nguyen AT, Cameron RA, Davidson EH. Developmental gene regulatory network architecture across 500 million years of echinoderm evolution. Proc Natl Acad Sci U S A. 2003 Nov 11; 100(23): 13356-61. Epub 2003 Oct 31.

    Amore G, Yavrouian RG, Peterson KJ, Ransick A, McClay DR, Davidson EH. Spdeadringer, a sea urchin embryo gene required separately in skeletogenic and oral ectoderm gene regulatory networks. Dev Biol. 2003 Sep 1;261(1):55-81.

    Davidson EH, Rast JP, Oliveri P, Ransick A, Calestani C, Yuh CH, Minokawa T, Amore G, Hinman V, Arenas-Mena C, Otim O, Brown CT, Livi CB, Lee PY, Revilla R, Schilstra MJ, Clarke PJ, Rust AG, Pan Z, Arnone MI, Rowen L, Cameron RA, McClay DR, Hood L, Bolouri H. A provisional regulatory gene network for specification of endomesoderm in the sea urchin embryo. Dev Biol. 2002 Jun 1;246(1):162-90.

    Davidson EH, Rast JP, Oliveri P, Ransick A, Calestani C, Yuh CH, Minokawa T, Amore G, Hinman V, Arenas-Mena C, Otim O, Brown CT, Livi CB, Lee PY, Revilla R, Rust AG, Pan Z, Schilstra MJ, Clarke PJ, Arnone MI, Rowen L, Cameron RA, McClay DR, Hood L, Bolouri H. A genomic regulatory network for development. Science. 2002 Mar 1;295(5560):1669-78.
    Last edited: Apr 28, 2008
  7. Apr 28, 2008 #6
    I think the real reason that people who work on organ systems don't usually call what they are doing systems biology is not because the name isn't approriate as moonbear said. instead it is because there tends to be a more appropriate name at that scale i.e. endocrinology, cardiology, renal physiology etc.

    Whereas people studying subcellular systems would have to call themselves "metabolomicists", "proteomicists", "phosphorylomicist" and other assorted silly sounding names. "Systems biologist" sounds much better.

    Also, JonMoulton I read that first science paper you posted from 2002; it's very interesting stuff.
  8. Apr 29, 2008 #7

    While this may be the state of systems biology today (it is still in its infancy), I believe the overall goal is to be able to apply this on broader scales such as organs and organisms. The individual networks need to first be fully understood before they are joined with larger scale and more encompassing networks.

    I believe the ultimate "wet dream" of systems biology is to eventually be able to understand all these networks and how they all interact in such detail to one day have a computer program that simulates a model organism (say a model human). So that when certain perturbations are made to the program it can accurately and reliably predict certain outcomes.
  9. Apr 29, 2008 #8
  10. Apr 29, 2008 #9
  11. May 1, 2008 #10
    At this moment systems biology is often used as a buzzword to fund mostly "omics" experiments. However one of the stated goals is the (mathematical) modelling of biological systems. Usually this requires biological high-throughput experiments to generate the data that actually allows the creation of suitable model.
    However, depending on the field that tackles that question, and often also depending on country and conference the immediate goals can vary.
  12. May 2, 2008 #11


    User Avatar
    Staff Emeritus
    Science Advisor
    Gold Member

    It actually gets more specific than that. It really is just the latest buzzword to describe what some of us have been doing for quite a long time now (and what some people I know have been doing for at least the past 30 years). For example, I study reproductive biology. My approach fits with the systems biology approach, but I've been doing it long before the term systems biology was coined (as people were getting into genomics and proteomics, I was joking that eventually they'll come back around to realizing they need to understand functionomics...how it actually works in the whole organism, not just single cells). I've studied at different times in my career reproductive behavior, endocrinology, neuroendocrinology, neuroanatomy, neurotransmitters and their receptors and transcription factors within the brain and pituitary, and a little bit of ovarian biology, as well as a bit of interaction between nutrition and reproduction. I've studied a variety of organisms to understand the functions of some of these parts of the system, and my overall focus is on understanding how ovulation and reproductive behaviors associated with ovulation are synchronized. I don't do this all on my own, as I have collaborators who have their own areas of expertise within the field, but we work together to understand the whole system, not just how a particular gene or protein functions in a vaccum. This requires using quite a range of techniques from molecular to whole animal (everything from qRT-PCR, in situ hybridization, immunocytochemistry, pharmacology, radioimmunoassay and ELISA, to simple behavioral observation...I haven't used electrophysiology, but others in the field who I know well are doing those studies and publishing the relevant findings in that area). But that's the point, you don't let your techniques or animal or other model limit you, you ask the questions that need to be asked about the system you're studying, and then seek out the right collaborators to help you set up whatever you need to do to tackle the next question.

    This is part of why I always have difficulty explaining to people what my area of expertise is...I'm not JUST a physiologist or neurobiologist or animal scientist or cell biologist, I'm a reproductive neuroendocrinologist. That's the system I study, and doing so bridges many traditional fields/departments.
  13. May 5, 2008 #12
    I agree partially with that statement. One of the ultimate goals of systems biology, however is not only to biologically understand cells (or biological entities of any other level) as we are used to do, but to quantify it in such a way that mathematical modeling becomes possible. The idea is to be able to do with biological systems what theoretical physicists are doing with theirs. It is a very ambitious idea and, in my opinion, we are still not able to generate data of the required quality. This is one of the reasons why systems biology has become an empty buzz-word. There is as of yet no real consensus how to get to that point. One difference to what has been done in the last 30 years or so is the increase of of data throughput (and the whole new field of data analysis and interpretation that has or is being implemented to deal with the data) during the last decade or so. It has opened up some new avenues, but as of yet, we are still away from reliable models.
  14. May 5, 2008 #13
    REALLY big and fast computers and MOUNTAINS of data from ALL pathways. :O

    The ultimate goal is not going to be obtainable anytime soon, but I suspect much will be learned along the way...
  15. May 5, 2008 #14
    The problem is not only the amount of data (the ability to create data now far outweighs the possibilities and approaches to analyze them) which, in theory might be addressed by increased processing power, but also the quality is an issue. One thing, for instance is temporal resolution. It usually takes minutes to process cells in a certain way before analytes can be measured. But many significant process occur within this time frame. Another point is e.g. resolution. Almost all analyzes are made on the population level, that is you work with whole tissues or other cell ensembles. However, no two cells (even if selecting for synchronized cells via cell sorters or using chemostats) are precisely in the same state. The results are always population-averaged. Now, if you imagine bimodal cell answers, you can easily imagine that this information is lost if you average the values across all cells.
  16. May 6, 2008 #15
    Sometimes there's more information in population measures than expected. For instance, the BOLD signal in a single voxel of fMRI data is correlated with the neural activity from a quite large area (order of millimeters). Jack Gallant at Berkeley showed a few months ago (in a Nature paper) that you can classify viewed images based on fMRI data from V1.

    Before that experiment, I would have thought you would be able to classify (or even reconstruct) images from V1, if you had single cell recordings from a (very) large sample of the relevant cells; this would be impossible in practice.

    I never would have guessed that the spatial organization of the receptive fields in V1 would be strong enough to produce the local signal bias necessary for such a course measurement as fMRI to be used for classification.

    The paper was: Identifying natural images from human brain activity by Kay, Naselaris, Prenger and Gallant
    Last edited: May 6, 2008
  17. May 7, 2008 #16

    Andy Resnick

    User Avatar
    Science Advisor
    Education Advisor

    There's tons of data. For example, there are fairly detailed and comprehensive models of ion transport and fluid flow through the proximal tubule of kidneys. The problem is that these models are not *predictive*. They all have many arbitrary parameters requiring experimental measurement for assignment (to draw another parallel with a particular physics theory) and are thus not useful. There's Cell ML (and others) out there as well- again, lots of adjustable parameters, no predictability.

    The real problem, IMO, is the lack of a fundamental theoretical principle. Physics has conservation laws and thermodynamics; Chemistry has atoms and molecules; Biology has nothing yet. Even the most fundamental 'laws' in biology (gene -> protein) has counterexamples (catalytic RNA). Find me a counterexample to the conservation of energy.
  18. May 7, 2008 #17
    That is basically what I am trying to convey. The data is not good enough to allow modeling precise enough to have predictive powers, at least not for . I kind of disagree that biology has no fundamental principles per se, because all principles you listed from the field of chemistry and physics apply to biological systems. But they are simply too complex to be described within a simple algorithm. Hence there are models to approximate them. I am not saying that there are no models. There are plenty of them. Some even with predictive powers (I recall a study that had modeled glycolysis based on metabolite data that were measured with seconds or ms in temporal resolution and could predict that the proposed pathway based on genome data was inomplete). However, systems biology tries to do it essentially for a whole cell.
  19. May 11, 2008 #18
    Well, it seems that most of biologists don't know about cybernetics, systems theory, control theory, complex adaptive systems theory or Nonlinear Dynamics. Nevertheless the author of the "General Systems Theory" is Ludwig von Bertalanffy (a biologist). As Olaf Wolkenhauer say: "Systems Biology is a merger of Systems and Control Theory with Molecular and Cell Biology". Ideally "Systems Biology" implies a theoretical work (mathematical and computational) supplemented with experimental work (quantitative experiments).
    Control Theory is generally given since the undergraduate level in Electrical Engineering, Chemical Engineering and Mechatronic Engineering programs.
    Perhaps you should know a little about concepts related to "Systems Science" to understand the "Systems Biology" approach. Or mainly differentiate the systems approach from the reductionist approach.
    Systems Biology generally study biological networks like signaling pathways, regulatory networks and metabolic pathways.
    But there is some efforts tu develop a model of a whole cell, like the "Leading Project for Biosimulation" in Japan.
    And also some efforts to develop a model of a whole organ like the "Physiome Project".

    I invite you to visit my blog: www(dot)systems-biology-research(dot)blogspot(dot)com

    Well I like the systems biology methodology or philosophy but I'm not sure that quantitative information like related to protein concentrations, coul be sufficient to describe a biological phenomena. I think that communication between biomolecules is through signals, and that the physical nature of this type of signal isn't known yet. The nature of this signal could be perhaps electromagnetic. I haven't found a Lab that do research about that. And it is directly related with the physical nature of the soul. I think in that way because I'm Electronics Engineer, and in my career I study in general systems that process information.
    Well, if you think I'm wrong or that I'm crazy, I'll understand.
    After all I'm only Electronics Engineer, I don't have still a PhD degree, I'm only an engineer.

    At first I desire to specialize in Systems & Synthetic Biology. But if I could I would like to study the physics of biological communication.

    I know that at the moment the scientific community think that two structurally matching biomolecular objects exchange specific information by mere contact, but I think that is wrong.
  20. May 11, 2008 #19
    Try these (one example each from quantum and classical theory):


    http://www.cheniere.org/techpapers/...olation of the Conservation of Energy Law.doc

    Also, to your second statement about biology---there are theory/laws in biology:

    1. Cell theory...

    2. Theory of evolution, for a long time now...

    3. Hardy-Weinberg Law...
    http://www.iscid.org/encyclopedia/Hardy-Weinberg_Law [Broken]

    4. Scaling Law:

    A law describes what nature does under certain conditions, and will predict what will happen as long as those conditions are met.

    A theory explains how nature works

    Biology, just as physics and chemistry, has both.
    Last edited by a moderator: May 3, 2017
  21. May 19, 2008 #20
    Several of the comments above were along the lines of:

    "Biology is too complicated to come up with predictive computational models given what we know."

    It is of course true that our understanding of the systems we are trying to model is incomplete; otherwise why would we be trying to model them? Certainly if there were some system in which we understood all the relevant components and parameters we could build a model that would accurately mirror the biological system.

    I submit that this is only one goal of theoretical/computational work and not the most important one. Having such a model would be nice for two reasons.

    1) You could easily use it to do in silico experiments and make predictions about emergent properties or higher order interactions between components which could then be tested empirically.

    2). If the model is predictive, then you would know that you have chosen to include the right components. That is, you would know what are the relevant components of the system in question by eliminating components one by one and seeing which system-level behaviors are preserved.

    These are perhaps in general form, answers to the two main questions/goals of systems biology. So the emphasis on finding ultra-predictive models is not wrong.

    However, there is another theoretical approach to solving the second of these questions that may be more tractable. Rather than building a single ultra-complicated / highly predictive model incorporating everything known about a particular system, we can build smaller-scale simpler models which incorporate very few details far more simply. Such models can answer the question of "what are the relevant components of a system" more easily than the large scale models which typically include (not obviously) unnecessary details.

    This has to happen by hypothesis-driven research. Theoretical biologists hypothesize that some particular component is the important one for generating a particular behavior (not all the behaviors of system). Then the simple model is built to determine if that component is sufficient for the behavior. Such a model would probably not look very predictive at all in the sense of reproducing experimental data. However, it is sufficient to test for some particular aspect of a system. For example, some variable may oscillate with high frequency oscillations on top of a low frequency oscillation. These models may reproduce just one of the oscillation frequencies.

    Such a model is sufficient for testing appropriate hypotheses concerning the sufficiency of a particular component to give rise to a particular behavior. If the answer is yes, that the component is sufficient, then little has been learned, the hypothesized mechanism is said to be strengthened somewhat by computational/theoretical evidence.

    But if the answer is no, then something very important has been learned. Showing that a component of a system that could plausibly cause some system-level behavior actually cannot possibly cause this behavior is an important result. This tells us that either we have not picked the right components of the system to investigate or that the behavior is an emergent property.

    In this way, we can investigate systems biology phenomena without building that one ultra-detailed/predictive model.
  22. May 19, 2008 #21
    Well, this is pretty much what is done in "regular" biology. I actually do not know any biologists that would call themselves theoreticians or theoretical biologists. Mostly this is done by bioinformatic guys. There have been quite a number of reports in which simple models (e.g. glycolysis) based on omics data have been created which had predictive properties. In one example the model based on flux data was not consistent with the proposed glycolysis pathway based on genome data. Subsequent analyzes resulted in the identification of an additional enzyme involved in metabolizing sugars.
    This is kind of part of systems biology. However, the real intention (and probably one of the reasons why it became a buzzword) is to be able to create such "ultra-detailed" models.
    In fact, it does not really have to be detailed per se, but it will probably be ultra-complex. One of the question is how much of the observed processes in the whole cell can be neglected or simplified and still result in a viable model.
    There are quite a number of efforts (again, bioinformaticians and mathematicians) to use rather simple models to predict metabolic flows within cells, with quite some success in some organisms (mostly E. coli, though).
  23. May 19, 2008 #22
    Yeah, that sounds right. My point was just that theoreticians can do it too.

    Maybe as you suggest, the difference with systems biology is the attempt to break away from explicitly hypothesis driven research. For example, the human genome project had no particular hypothesis, more of a "sequence everything and see what comes out" type approach. Perhaps the analogous research program for theoretical modeling is to incorporate every known detail into a model and "see what comes out".

    This may be the essence of the systems biology approach. If so, I am not convinced it is the right way to go...

    I recall a recent opinion article in one of those genome biology journals by Greg Petsko in which he argued that the genomics approach is great for studying the genome, but probably not the best way to study other things. I believe (not positive) that his area of interest is protein biochemistry...

    Here's the reference: http://genomebiology.com/2007/8/1/101
    Last edited: May 19, 2008
  24. May 20, 2008 #23
    Having worked at the borderline of "traditional" and "systems" biology I tend to agree somewhat with your sentiments. Too often (especially on congresses) systems biology is getting sold as something in the line of:
    biological data --> magic via bioinformatics -----> working model
    Only in few instances (at least that I know of) did this approach actually produce viable results. The interesting from Petsko is an interesting appeal not to use "omics" approaches on every biological discipline, which I clearly agree to. However, beside genomics there are other "omics" approaches that by their very principle are similar to genomics (hence the name) and are often collectively termed postgenomic techniques (incidently a field on which I worked on up until now). A quite recent and quite recent example is this paper

    Baerenfaller et al 2008, Science
    sciencemag.org/cgi/content/abstract/320/5878/938 (I cannot post links yet)

    The overall enthusiasm for this approach has now somewhat faded compared to the time when I started working on it (second half of the nineties or so) however data generated that way has proven to be useful in generating new hypotheses (but failed to be the magic bullet that one hoped them to be). However, there is still a conceptual jump from omics to systems bology (that is why I am usually saying that I am working on the omics or postgenomics field rather than doing systems biology, as some of my former colleagues claim). And to my knowledge there is still not obvious way to close this gap.
  25. May 20, 2008 #24
    I doubt all the hard-working bioinformatics and computational biologists would refer to what they do as "magic". :)
  26. May 21, 2008 #25
    Well, until it has been demonstrated that there is a viable way, I am going to refer to it as magic (or mathemagic). In fact, some of the bioinformaticians that I worked with have adopted that phrase...
    I do have to add that I worked with two bioinformatics teams, one of which was part of the same group in which I made my phd (it was a biological group with an adjunct bioinformatics subgroup). So having to have to deal with them so much I fell I am kind of allowed to make fun of them as much as I like (and vice versa, of course).

    Truth be told, however, it is mostly the biologists that think that the bioinformaticians got the solution to all their problems, so they kind of hope for that magic.

    A funny bit that I recall is what a Prof (initially a mathematician) referred to as ABC of modelling. A: assume data. I never heard what B and C were, as that one always cracked me up.
    But it refers to a basic problem, the data that biologists provide are often not good enough for the informaticians to base their modeling upon. Therefore often ad-hoc assumptions are made (the assume data part), which often are not very useful. In a way biologists and bioinformaticians (at least in these fields) often blame each other. The biologists want the bioinformaticians to provide an easy way to handle the data and use it to make meaningful models, bioinformaticians want better data.

    Also the same Prof told his students not to talk with biologists about their projects. They will only spoil the fun. Which is kind of true (I did spoil some phd projects, interestingly experimental data did not convince them that they were on the wrong track as much as some modeling done by myself).
Share this great discussion with others via Reddit, Google+, Twitter, or Facebook