Summary | Rationale | Powerpoint |
JPG Slide | JPG Image | Credits | Comments
| All Nuggets
Computational Approaches to Modeling the Brain and the Cell
Theoretical computer science can provide powerful tools to help understand how complex biological systems, such as brains and cells, process and exchange information.
Summary
The scientific understanding of complex systems typically evolves by a
process of iterative refinement in which simple models are replaced by
more complex ones. Historically this process of refinement has usually
proceeded in a rather ad hoc way. Now, however, a systematic approach is
called for as today's scientists strive for a comprehensive understanding
of highly complex information processing systems such as the human brain
or the intricate network of regulatory signals that are transmitted within
and between cells. Insights from theoretical computer science (TCS) can help
guide scientific investigation of these complex systems, similar to how
TCS insights and algorithms contributed to the Human Genome Project.
As one example of how TCS may aid scientific investigation in the future,
insights from the field of
computational learning theory could help guide the adaptive design of
experiments to uncover the structure of biological information-processing
systems that are currently not well understood. A large body of research
in this area studies the issue of what are the "right" questions to ask
(i.e. experiments to perform) in order to maximize the useful knowledge
that can be obtained about a unknown information-processing system.
Collaboration between theoretical computer scientists who study adaptive
learning techniques and biologists has the potential to greatly improve
the efficiency of scientific investigation.
Rationale
Put a slightly more detailed explanation here, which can be aimed at a general computer scientists.
Contributors and Credits
Scott Aaronson, Bhaskar DasGupta, Richard Karp, Rocco Servedio, Diane Souvaine
Image Ideas
Something like a visual representation of two figures, a computer
scientist and a biologist, with arrows going back and forth to represent a
flow of information.
See http://www.columbia.edu/~ras2105/temp/Presentation1.ppt
for an example (thanks to Bhaskar)
Comments
- to give feedback on this nugget, just add another bullet to this list
- I think the nugget may emphasize more the modeling challenge. A good model for the cell is essential for comparing, searching and prioritizing research, and is largely missing today. One may argue for instance that the success of the DNA sequencing effort is due the ability to model DNA as a sequence of letters. The cell of course is not so simple, it must be modeled as a complex information system. Thus, TCS may contribute...
- Along the same lines as the previous (anonymous) comment, the summary may give the impression that adaptive learning algorithms are the main/only way that TCS can contribute to this effort, as opposed to it being just an example. If the potential for contributions here is broader, then we should make that clear. - Salil
- More comments: Maybe the tagline could be a bit more catchy? The first sentence of the 2nd par of the summary may be a little too self-promoting, perhaps this can be conveyed in a more tactful way? - Salil
- Well written nugget. But the title promises more than the text delivers. Modeling experiments as an algorithmic process that can be optimized using TCS tools is a great point. But the title makes one believe TCS can also help explain the results of these experiments (and not simply design them). This is a tricky issue because we must sound bold yet credible. Neuroscience is a huge field and we must avoid creating the impression that TCS will model the brain. Rather, TCS will provide tools to help scientists "make sense" of whatever computational processes they've uncovered. -Bernard
- Revised nugget to try to address all these useful suggestions. -- Rocco
- It would be useful to have a concrete/concise example to illustrate how learning theory has helped to model (or "make sense" of discoveries) of a complex system. The current writing "Two hallmarks of successful CS theory research -- understanding layers of abstraction and dealing with models at the right level of granularity" is maybe somewhat intangible. Overall I feel a little more concrete evidence would make this a great nugget. -- Lisa
- I took out the somewhat intangible last paragraph, and added very brief mention of a concrete example in similar spirit (algorithmic contributions to the human genome project) at the end of the 1st paragraph. -- Rocco
- Added "(TCS)" in parens so that this acronym is defined. - Salil
- Note from designer Elaine Park: I wanted very much to bring the original idea to life, but I wasn't sure how to represent a computer scientist unequivocally without resorting to labels (which I think would interfere with the more important verbage). So here, data is being examined in a petri dish, which suggests simultaneously that biologist's research is being augmented by computer science, as well as the idea that the biologist is helping to culture computer science techniques.
- On first viewing, I didn't realize that these were petri dishes in the image, and thus the point was lost on me. Perhaps it's because the green sphere with 0's and 1's does not look biological enough... - Salil
- I did see them as petri dishes. The green contents of the dish with 0's and 1's does look spherical, which is a little weird (don't petri dishes hold basically flat gels? I don't really know). I'm not sure how it could be represented differently, though, and I do think the image works as it is. -- Rocco
- I would have liked to have a small colored circuitry somewhere in the picture to indicate we are looking at information exchange also, but not sure how and where to put it -- Bhaskar
- One could use test tubes instead of petri dishes or both. Circuitry with zeros and ones would be good. - Richard
- text is still too vague to make much sense. the relevance with the human genome seems tenuous at best. needs a concrete example or success story. (eg, PL people like tom henzinger have used PL techniques for modeling biological organisms). i like the picture. - Bernard
- Revised ppt uploaded. - Salil
- Note from designer Elaine Park: Following the suggestion, a little brain circuit is being grown in a petri dish. Plus a flask on the side for a little extra context.
