Computational Biology and Media Computation from Lee Mondshein

I taught a workshop on Media Computation in Waltham, MA this last July, which a Computational Biologist (Lee Mondshein) attended. He was pretty excited about Media Computation as an approach. I'm sharing his note to me here.

TO: Mark Guzdial

FROM: Lee Mondshein
SUBJ: Computational Biology and Media Computation: A Lively New Medium

DATE: July 7, 2008





I would like to share some thoughts that I have about a media computation approach to computational biology. At the very outset, I ought to note what I have in mind when I use the (rather flexible) term "computational biology." I employ this term to describe the use of mathematical and computational concepts and techniques to explore, analyze, model, and elucidate the dynamic processes of living systems. For me, the terms "computational biology" and "systems biology" refer to much the same activities, but with a shift to a broader scope in the latter term. My long-time interests as a computer scientist focus on the (now popular) biological concern with dynamic biological activity rather than static biological structure.



Question 1. What Biological Focus ?

If pictures and sounds are fundamental phenomena in the domain of media computing, what would be the corresponding phenomena in a biologically-focused application of the media-computation teaching/ learning paradigm ?



Response.

Pictures can be represented as data. Sounds can be represented as data. Biological processes can be represented as data (in many ways) — and such data can be manipulated in a variety of ways ...

For example, in recent years, considerable work has been done in representing cellular processes as modular processes using nested algebraic notation (see Regev et. al.; work on process algebras - pi calculus and its refinements). The "collage" ideas developed in media computation can be generalized to these modular processes, via a composition operation that is native to process algebras.



Question 2. What Sensate Result of Computational Activity?

In applying a media computation perspective, what would the "sensory" analogy of pictures and sounds, in the realm of computational biology?



Response.



Visualizable algorithms.

This question has an easy version and a hard version. The easy version arises from easing off the focus on direct representation of processes, and looking instead at secondary data derived from the super-abundant experimental observations of such process (e.g., via so-called microarrays, which can detect, for example, what genes are actively related to what conditions (of health, of environment, etc.) ). These sets of data can be analyzed using a variety of pattern recognition/ clustering algorithms (with lots of parameters to tweak); the varying results can be displayed and explored in striking visual form. These results can lead to fresh questions, and new experiments and data analyses. There is ample media-computational "low hanging fruit" to be harvested in this visual-sensory territory.



A Truly Lively Medium.
The "hard version" of the question has to do with pushing the envelope, and exploring the ultimate visual correlate of the algebraic process representations that I mentioned above— namely a resulting life-form. One could attempt to design (idealized, highly simplified) algebraic representations of the process of gestation and development, based on a modular "collage" of genes, gene-expression products, and associated feedback control processes. Thanks to the underlying computational theory, these processes would be executable. The observable, sensate output would be the appearance of the creature itself. Now there's a lively medium ! A tweaking of the algebraic process-data would have the stunning effect of changing the life-form ! One could explore analogies with image and sound processing algorithms (e.g, alternative factorings of computational processes), as well as analogies with computational architectures (e.g., "aspect-oriented" computational phenomena possibly embodied within the raw biological substrate). I have lots of ideas.





I would be interested in (incrementally) exploring some of these and other ideas with you. Some are at the gestational stage; others are quite mature. I have been teaching computational biology and bioinformatics -related courses for over five years (to graduate and undergraduate students). Much of my course content has been shaped by the preceding twenty-odd years I spent at MIT Lincoln Lab, Draper Lab, and MITRE Corporation working on a variety of speech, radar, and image processing projects.



I am finding your workshop tremendously stimulating. As you can tell, I am very excited to gain a first-hand appreciation of the meaning and benefits of the media computation paradigm. I am especially impressed by the energy of engagement that results from this approach, by the exposure it provides to ideas about data-polymorphic algorithms, and by its adaptability to a variety of subject domains.





Best regards,
Lee Mondshein



Mathematics and Computer Science Department

Bridgewater State College

Bridgewater, Massachusetts