sticklen@CPSWH.CPS.MSU.EDU (Jon Sticklen) (05/23/89)
I agree with Peter Karp's note that it would be useful to have a clarification of the goals of the biomatrix project. I suppose Peter and I, as both CS folks who are following the biomatrix work with interest, would be helped to know how we can contribute by knowing better what you all want. To address Peter's comment, it would seem that there are some real differences between deep and shallow expert systems, but to pursue that here in detail would not too productive - its an argument mostly rooted in technical AI issues. I will look for Peter's paper with interest. But there is one facet of Peter's note that I think is important for discussion here. Peter raised the issue of "surface level" understanding being an approximation to "deep level" understanding that is sometimes practically speaking the only way we can go; as in - we have to model the celestial mechanics of the solar system by approximation techniques even though we know that an exact solution is theoretically possible. The general model that Peter is relying on here is one in which one can think of layers of description of some phenomena with the lowest level being the "first principles" level and each successively higher level being implemented on the next lower one: ie, a reductionist model of understanding. A difficulty may arise when we try to apply this model to biological phenomena as follows. When we try to reduce one level of understanding to another, we loose the descriptive power that the terms at the higher level afforded, and are forced to use the vocabulary at the lower level. Eg, mating behavior of large mammals may be reduced through a number of steps (in principle perhaps) to the bio-molecular level and beyond. But if we do that, then we cannot use terms like "drive to find a mate" because such terms do not exist at the bio-molecular level. One of the most pressing needs when trying to represent a phenomena is to search for the appropriate level of description to capture understanding. Going further, one of the most pressing needs of the bio-matrix project (it seems to me) is to find a representation scheme that is robust enough to support descriptions of different phenomena at many different level of abstraction, and that allows each description to utilize terms appropriate to the phenomena being described. Jon Sticklen AI/KBS Group - CPS Dept Michigan State University East Lansing
chiafari@umbc3.UMBC.EDU (Mr. Frank Chiafari ) (06/01/89)
Gentleman, I have been reading bio-matrix news for a long time and feel a responsibilty to answer dave's request for input from the biological community The idea of a bio-matrix being built to inter-relate data between organizational levels in biology may be premature. It concerns me that biology as a field has not developed sufficently to enable rational relationships to be derived from between disiplines. I also do not believe that the historicallyaccepted divisions of biology represent hierarchical levels of information which should be integrated to effect data interpretation. An example of where such an integration would have been disastrous is in the application of homeotic sequences from their original finding in drosophila to higher verts. I remember a number of embryologists and ecologists claiming that any homology found couldn't possibly have developmental implications!! An organization of ideas, by definition, limits the choices available. There exists, however, a need to apply computers to the problems of database managment and to build up from data gained at the molecular level to applications in system organization.. I have attempted such a project, but within the same level, taking sequences recognized by promoters and trying to determine a consensus sequence and recognition model. I ran into a number of problems: 1) The question I was asking required many computer hours and a working understanding of computers so that I could program the machine myself when building new models. Few programers were willing to spend the time necessary to learn enough biology to understand the problem. 2) The results I obtained were not readily accepted by the community, and they may never have been published if experimental results had not "caught-up" with my predictions. 3) Specific to sequence analysis, the literature is relatively inconsise as to how to decide the signifigance of a result. This may reflect my own failing rather than a shortcomming of the field. In summary, I offer concern, NOT condemnation With Regards Francis A. Chiafari