threaded | unthreaded | newest first

Predictive in silico modeling of an entire organism

topic posted Wed, October 3, 2007 - 7:51 PM by  Randy 		Share/Save/Bookmark
Advertisement
How soon do you think we'll be able to model an entire organism using computers, if ever?

I personally believe it will be soon, a matter of a few years at most.

I'm a layman interested in the subject, and have started a Google groups page devoted to general interest literature that is tangentially related to modeling an organism, or the "virtual worm, weed, and bug", to borrow a phrase from one of the files posted on the groups page.

Here's a link to the page: groups.google.com/group/bebobio

Let me know what you think. Thanks! Randy
posted by:
Randy
offline Randy
Seattle
2 friends
join to post
Advertisement
Advertisement
  • Ben
    Ben
    offline 37

    Re: Predictive in silico modeling of an entire organism

    Sun, February 17, 2008 - 2:41 AM
    Actually I don't think it'll be anythime soon.
    Like bordering on never.
    Have you taken a look at an individual cell?
    It's insanely mind-boggling how complex just that is.
    join to post
    • Randy
      Randy
      offline 2

      Re: Predictive in silico modeling of an entire organism

      Thu, February 28, 2008 - 8:52 PM
      Thanks for posting! You may be correct, but I don't think so. Here's why I believe it's possible.

      A cell is robust.
      A hallmark (definition?) of robustness is simplicity.
      Therefore complexity in a cell must come from a network of simple reactions.
      Each simple reaction can be modeled.
      All we really need to model is the web of relationships between individual simple reactions.
      I believe that will be possible soon, and that in retrospect it will be obvious.

      PS If you're posting at 3:00 in the morning on a Sunday, you must be a party animal extrodinaire! I'm impressed!
      join to post
      • Unsu...
         

        Re: Predictive in silico modeling of an entire organism

        Fri, February 29, 2008 - 8:20 AM
        The reductio ad absurdum counterpoint:
        With that reasoning and a basic understanding of physics, which is also ultimately a web of relationships between individual simple reactions, we should be able to model *any* physical system. But a 100% accurate weather model, let alone Assimov's Foundation are a ways off yet. (And that's not even addressing Heisenberg nondeterminism)

        Chemistry is simple. The interactions of a small number of, say, peptides is simple. But things get exponentially harder as you add more elements to the system. And then when you abstract certain reactions to their symbolic equivalents (a way to do a sort of in-model data compression) you lose certain subtleties of detail. Of course there is going to be a happy medium somewhere between a Tim Conway representation of a cell (alive or dead) and an atom-for-atom model; Progress is a matter of pushing the model from one end of the scale to the other.
        join to post
        • Randy
          Randy
          offline 2

          Re: Predictive in silico modeling of an entire organism

          Sat, March 1, 2008 - 4:39 PM
          Spiny, thanks for your response! I agree. My argument, when taken to extreme, leads to an untenable determinism. I also agree that the type of modeling I’m interested in falls further down the scale from an atom-by-atom exact model. Here are a few thoughts regarding the comparison of modeling the weather with modeling an organism in the context of the points you raise.

          I can’t put my finger on it, but there seems to be a fundamental difference between the rules the weather follows and the rules of biology. Obviously both start with the laws of physics, but biology has additional constraints. Let’s say you start with a weather system in a stable, uniform state and add some minuscule perturbation. If you were able to do the “same” experiment several times, chaos theory shows that the results would vary widely. Now look at protein folding. In nature the process is repeated untold times under all sorts of variance, yet life relies on getting essentially the same result every time. I could use some help from someone more knowledgeable to characterize the difference between these two types of complex systems.

          In developing a model of a biological system, we have testability and reproducibility on our side. If you were to make an in silico prediction for a particular mutation in a bacterial cell, it would be no problem to make the mutation and test your prediction a million times in a small flask.

          As I’m sure you are aware, the piles of data generated by modern mol-bio are huge and only getting bigger. Although human beings are excellent at pattern finding, I think it’s unlikely any human will ever be able to synthesize the data to understand an entire organism and make useful predictions. The computer will be the tool that enables us to do it.

          I appreciate your perceptive comments, and would like to ask you a favor. If you have a chance, take a look at the BeBoBio Google Groups page groups.google.com/group/bebobio
          The page is dedicated to files, book reviews, and discussion of biomodeling.

          Foundation! It’s been about 30 years since I read that one. Have you read any Greg Bear, especially “Blood Music”?
          join to post
          • Randy
            Randy
            offline 2

            Re: Predictive in silico modeling of an entire organism

            Mon, March 10, 2008 - 7:19 PM
            From the website of Garrett Odell, University of Washington, Director of the Center for Cell Dynamics

            www.celldynamics.org/celldyn...dex.html

            "The real biological facts are arriving in overwhelming abundance. We can change the genomes of study organisms easily and quickly and observe the emergent consequences. Computers are now astonishingly fast and cheap, with vast storage capacity. Computer languages such as Java are up to the task of translating mathematical formalities into object-oriented code that works. New computer-controlled optics and new fluorescent probes let us see in live cells what cell-level actions actually emerge from the self-organizing molecules that genes encode - let us see what actions should therefore emerge from formal models based on classic dynamical systems theory. We have in hand, somewhere in the reductionist fact avalanche, solid foundation blocks for those models. We have the computer power to connect one to the other. Just possibly, combining these tools and empirical discoveries, the generation of scientists now in their 18's - 30's will be the first to conceive and popularize a creation myth as sublimely beautiful as the ancient ones, but differing from them by being true. If you could participate in that, what beckoning calls from other choices of career or amusement would even be audible?

            It's still early days on a very long quest. It will take the life's work of thousands of scientists working across the world to do it. We aspire here at the CCD only to contribute. If it succeeds, the side effect NIH hopes for will be that we will come to understand complex biological systems so deeply that we can fix them when they break. Another consequence will be that this generation's longest-remembered achievement will be that, collectively, we figured it out."
            join to post
            • Randy
              Randy
              offline 2

              Re: Predictive in silico modeling of an entire organism

              Tue, March 11, 2008 - 9:38 PM
              Here's another one from NRCAM National Resource for Cell Analysis and Modeling

              www.nrcam.uchc.edu/about_nr...RCAM.html

              Approaches in computational cell biology are coupled with high resolution light microscopy to facilitate the interplay between experimental manipulation and computational simulation of specific cellular functions that can range from simple molecular motors to tissue-wide process. The Virtual Cell is deployed as a distributed application that is used over the Internet. It is freely accessible to all members of the scientific community.

              The Virtual Cell has been specifically designed to be a tool for a wide range of scientists, from experimental cell biologists to theoretical biophysicists. Likewise the creation of models can range from the simple, to evaluate hypotheses or to interpret experimental data, to complex multi-layered models used to probe the predicted behavior of complex, highly non-linear systems. Such models can be based on both experimental data and purely theoretical assumptions.
              join to post