Banner
    Systems Biology Has Become Meaningless
    By Michael White | January 27th 2010 10:57 AM | 13 comments | Print | E-mail | Track Comments
    About Michael

    Welcome to Adaptive Complexity, where I write about genomics, systems biology, evolution, and the connection between science and literature,

    ...

    View Michael's Profile
    Since when has systems biology been a synonym for genomics?

    This is from a Perspective piece in the Oct. 2 issue of Science:

    The relative value of discovery aimed at hypothesis generation versus hypothesis testing has been debated. High-profile journals publish systems biology studies, including the human genome sequence, but most papers focus on hypothesis-driven investigations.


    The human genome sequence is systems biology? Really? I can't blame the author here: most "systems biology" conferences I've attended have been indistinguishable from genomics conferences. For some, systems biology is just a sexier term for genomics. If we're measuring the entire 'system' at once, with the latest genomic technique, such as ChIP-seq of various histone forms, that's systems biology, right?

    Unfortunately, this kind of thing has made the term systems biology radioactive in certain segments of the biomedical community. Systems biology has become the latest buzz-word, instead of an emerging discipline focused on new questions that were difficult to study 20 years ago.

    Sequencing a genome is not, and should not be systems biology. Instead of Genomics 2.0 (which is basically, in my view, cartography: better and better high-throughput or computational techniques that map genome-wide phenomena of protein-protein and protein-DNA interactions, gene expression profiles, chromatin states, etc. - important, but still genomics), systems biology should be a natural evolution of genetics and biochemistry. Systems biology should be a hypothesis-driven field that studies the dynamical principles that account for the behavior of biological regulatory systems like gene circuits.

    Systems biology should build on the questions about regulatory systems that were being asked by biochemists back in the 1970's and 1980's - before people had some of the amazing computational and experimental tools we have now.

    This conflation of genomics with systems biology has had negative consequences for those of us who are interested in hypothesis-driven, principles-based questions about regulatory systems. If we attend conferences (or read papers) labeled systems biology, we're unlikely to find the kind of research we're looking for, which is instead found at meetings like "Computational Cell Biology."

    Even worse, in a typical genetics or biochemistry NIH study section, if your stuff gets labeled systems biology, you're toast, as I was told directly by one of the program officers at the NIH. Old-school biochemists, molecular biologists, and cell biologists, who should be interested in fruitful models regulatory systems, associate you with 'hypothesis-free' genomics/computational biology, and quickly write you off.

    For these people, the term systems biology has been tainted beyond rehabilitation, so perhaps we need a new term to capture the kind of work that's being done by people like John Tyson, James Ferrell, Michael Elowitz, Michael Savageau, Eric Siggia, and many, many others who are using both experimental and mathematical tools to understand the principles underlying the behavior of biological regulatory systems. Something like "Biological Dynamics" could capture what is distinct about this unamed field, distinguish this work from genomics, and help people recognize that there are some exciting questions that should be addressed, now that we have some very good maps of the interactions that are happening inside of the cell.

    Read the feed:

    Comments

    Hank
    Instead of Genomics 2.0 (which is basically, in my view, cartography ... systems biology should be a natural evolution of genetics and biochemistry. 
    That's as concise and lucid a perspective as any of are going to find.
    Sort of the way that "radioactive" is used here

    adaptivecomplexity
    Just to be clear (I don't think I was was lucid enough!), I'm not denigrating 'cartography'. I use expression, ChIP, interactome, etc. data all the time, and I'm glad it's available.
    Mike
    AdamRetchless
    Darn, I had gotten the impression that Systems Biology, being used in the names of institutes in Seattle and Boston, had become a respectable term. It's really sad to hear that it isn't even understandable.

    The contrast between "cartography" and an "emerging discipline focused on new questions" reminds me of an essay I read regarding the four stages of scientific research. Basically, these types of research would represent stages four (mature field) and two (emerging field), respectively. This type of confusion may be common, since new fields may often arise from other fields that have reached maturation.
    adaptivecomplexity
    Thanks for the link to the article. Systems biology is hot in some quarters and viewed with derision in others. The fact that it's ill-defined makes it really difficult to navigate.
    Mike
    I don't think anyone working in systems biology (SB) would agree that it's merely a synonym for genomics. SB is about integrating all -omics data types within the context of powerful computational methods, often to propose new testable hypotheses, not necessarily to test an existing hypothesis. For example, in the context of cancer research, SB approaches are tremendously useful for proposing new candidate gene markers, which of course have to be validated by traditional wet-bench methods before they become clinically useful markers.

    It's too bad the NIH holds this view, assuming they do. By contrast I heard in a talk by Atul Butte (few years ago) that the number of awarded grants with the word "informatics" in the title/abstract had risen sharply in the last 10 years. Informatics is often a code word for "SB approach inside".

    adaptivecomplexity
    SB is about integrating all -omics data types within the context of powerful computational methods, often to propose new testable hypotheses, not necessarily to test an existing hypothesis


    I still see this process as omics. (I suppose I have the bad habit of lumping all omics under the term genomics. Is doing ChIP-seq experiments on all transcription factors in yeast genomics or proteomics?) I see it as omics because, even if you're integrating various types of data together, it's still largely phenomenological, 'let's broadly survey the lay of the land' type of thing. I'm not criticizing it; I don't share the negative view expressed by some that omics is purposeless, hypothesis-free work. And like you say, it does often generate hypotheses. So I'm not criticizing it.

    But I don't see this as something fundamentally different from what genomics has been for years, before the term systems biology became popular. If this integrated omics really what systems biology is, then we need another term for the distinctly different science I'm describing as 'biological dynamics.' They are not the same thing.
    Mike
    I still see this process as omics. (I suppose I have the bad habit of lumping all omics under the term genomics. Is doing ChIP-seq experiments on all transcription factors in yeast genomics or proteomics?) I see it as omics because, even if you're integrating various types of data together, it's still largely phenomenological, 'let's broadly survey the lay of the land' type of thing. I'm not criticizing it; I don't share the negative view expressed by some that omics is purposeless, hypothesis-free work. And like you say, it does often generate hypotheses. So I'm not criticizing it.

    Understand now that you are not criticizing SB, it was the title of the post that threw me ;-)

    As to your question, I would say ChIP entails aspects of both prote- and gen-omics, but a ChIP assay in itself is not an example of an SB approach, it's a particular type of experiment with a relatively narrow mechanistic focus. Nothing wrong ChIP, mind you. In fact, the results of ChIP experiments and others feed databases which are critical to SB.

    I've had to address the "phenomenological" charge before. While it is true that an SB biologist is unlikely to make a conclusion of the form: "protein X is necessary for process Y", I don't understand why, for instance, showing by an integrated SB approach (genome + proteome + interactome + math ) that genes frequently mutated in human cancer tend to cluster together in the human interactome, is merely "phenomenological"? (Or maybe that's not the kind of example you had in mind). Such a finding can immediately impact drug discovery strategies as well as certain disciplines in translational medicine. If anything, I would say that GWAS studies, for instance, are more prone to the criticism of being phenomenological, because frequently they render no mechanistic insights, they merely "associate" with a particular phenotype. Frequently nobody knows why.

    I'm not too concerned about what we call SB, but it's not merely genomics under a new name.

    adaptivecomplexity
    I was being deliberately provocative in the title, and I wasn't as clear as I could have been when it comes to distinguish my views from those of critics of genomics or systems biology. Just to be clear, I'm not denigrating work I characterize as phenomenological, like the cancer genome atlas projects, etc. I've done work like this myself.
    But is finding genes frequently mutated in human cancers any less phenomenological than identifying the target genes of a transcriptional regulator by a ChIP-seq experiment? Maybe what I'm doing here is using phenomenological as a (non-derogatory!) substitute for non-hypothesis-directed research. 

    Contrast this with efforts to mathematically model the MAPK pathway in Xenpus oocytes, or the exit from mitosis, or the production of gene expression stripes in the the fly embryo. Both types of research could be considered the study of systems, but in fact the types of questions and the approaches used are significantly different - too different to be considered the same field, even though they get lumped together into one department at places like Harvard's Systems Biology Dept. (You can pick either one of their 2 sys. bio. depts.) At our university, the genetics/genomics program and the bioengineering program are arguing over who gets to use the term systems biology for their graduate program.  You can also see the differences if you contrast the journal Molecular Systems Biology with Biophysical Journal.

    The problem with reviewers is this: when you're modeling the cell cycle, your grant proposal goes to a study section that focuses on genetics or molecular or cell biology, and not on omics. And there is a significant portion of the non-omics community, as I'm sure you've encountered, that does view phenomenological, "hypothesis free" (see the article I linked to at the top of my post) research fairly negatively. Anything computational, especially if it's labeled systems biology, can easily lumped into this no-hypothesis category and dismissed.
    Mike
    adaptivecomplexity
    It's too bad the NIH holds this view
    In the instance I relate, the view was not held by the NIH officer, it was held by members of the study section, who of course are not NIH employees. And naturally there are some study sections and NIH institutes more sympathetic to omics and informatics than others.
    Mike
    I agree that the sequencing of the nucleotides in an entire genome and identifying genes does not represent "systems biology" any more than counting up all of the animals, plants and microbes in forest represents "ecology." Genome sequencing efforts to identify proteins does provide a valuable parts list, and with comparative genomics we can get a hint of the critical nature of many of these proteins across the species. I am afraid that the suggestion of "biological dynamics" as a new term to replace "systems biology" suffers from the same ambiguities. The term "systems biology" was always a poor descriptive of its intended usage to encompass the analysis of complex systems within the broad realm of "molecular biology." The problem with using "biology" in this context is that this can just as easily be taken to include the interactions of diverse organisms within ecosystems. While "systems biology" may be viewed as an area of investigation under "genomics", which is often unfortunately the case with granting agencies, it is at the protein level that real direct interactions are manifested that result in phenotypes. At Kinexus Bioinformatics Corporation, to be more specific about the study of these important protein interactions, be they with proteins or other molecules, we have coined the term "systems proteomics." It is only within the context of "proteomics" that these components can be assembled into "systems," whether describe for example macromolecular machines such as ribosomes, linear metabolic pathways or sophisticated protein kinase-based signalling networks. At the very least, the genomics sequencing efforts have reveals the existence of at least 23,000 human proteins. In view of the sheer number of proteins and their potential interactions, it is rather foolish to think that one can try to elucidate the operations of these proteins in a purely hypothesis-driven manner.

    Mark Changizi

    Agreed. For a while some years ago (when I was more theoretical-biological than now) I figured maybe I was a systems biologist, interested in the systems-level functioning of biology. I quickly learned I wasn't a "systems biologist."
    adaptivecomplexity
    So you know what I'm talking about. Some clarification in the naming would help, I think. My proposed term 'biological dynamics' is inspired by 'nonlinear dynamics', but I suppose my term could also come off vague.
    Mike