Report on the collaborative project 
between ICOT and the NIH 

Richard J. Feldmann 
Division of Computer Research and Technology 
National Institutes of Health 
Bethesda, Maryland 20892 

The collaborative project between ICOT and the NIH must be considered at sev-
eral levels. At the highest level the project is meant to provide a vehicle for developing 
friendship and understanding between Japanese and American scientific workers. By 
means of visits to each others laboratories, by almost daily fax and e-mail messages we 
have begun to understand each other's ways of thinking. 

Two specific scientific projects were used as the scientific substrate for the project: 
Genetic Information Processing and Protein Folding. 

The Genetic Information Processing work is very much influenced by the ICOT 
style of logic programming. This work is being done also in collaboration with workers 
at the Argonne National Laboratory (ANL) and the Lawrence Berkeley National Lab-
oratory (LBNL). Over the last year and a half, four workshops have been held. The 
emergence of the InterNet means that workers can come together in one physical place 
to meet and talk but still use the computers and databases in their own laboratories. 
A very graphical interface and database program, called GenoGraphics, has developed 
from these workshops. GenoGraphics which is the work of Ross Overbeek and Ray 
Hagstrom from ANL, started with the data representation of George Michaels (NIH) 
on the E. coli genome and the work of Kaoru Yoshida (ICOT and LBNL) and Cassan-
dra Smith (LBNL) on human chromosome 21. The E. coli data was collected by Ken 
Rudd who is now a member of the National Center for Biomedical Communication 
and Information (NCBI) in the National Library of Medicine (NLM) of the NIH. Dr. 
Michaels has just recently held such a workshop at the NIH. Workers from all over the 
USA and from England came together to increase the range of genomes which can be 
handled by GenoGraphics. During this workshop the genome for yeast pombe collected 
by the workers at the Imperial Cancer Research Fund (ICRF) laboratory in London, 
England was introduced into the GenoGraphics logic programming data format. The 
ICRF workers had spent almost half a year developing programs and organizing their 
data. During the week-long workshop they were successful in transferring their data 
to logic programming format. 
- 56 -

Dr. Michaels expects that in the next year there will be fragments of about 50 
genomes for various small organisms entered into the GenoGraphics format. Geno-
Graphics as a result of Dr. Hagstrom's work is now written in C and runs on any PC 
compatible machine. We expect that GenoGraphics will become a world standard tool 
for the representation, manipulation and investigation of genomes. 

The collaboratory model which George Michaels has developed is a powerful out-
come of our interaction with ICOT and the other US national laboratories. Scientific 
workers can now come together from all over the world and using the InterNet can 
work together effectively for a short period of time. 

In the Protein Folding portion of our collaboration with ICOT we have built and 
analyzed several models for the representation of protein structure. The collection and 
analysis of x-ray crystallographic data sets was begun in our laboratory almost 20 years 
ago. The relationships between protein structure, function and folding pathway have 
been very difficult to elucidate. The protein folding problem is the key technology 
which will enable biological system design. During the collaborative project workers in 
both countries engaged in the design and construction of both physical and computer 
models. Physical models provide simple, visual, trans-cultural vehicles for communica-
tion. Computer models can be constructed to represent salient features of the physical 
models. Using both logic programming and conventional machines we have investi-
gated the statistics and dynamics of these models. The resolution of a protein model 
and its water environment is a critical determinant of the computer power required to 
simulate folding. Parallel computational techniques for simulating protein folding using 
logic programming machines have been developed by our ICOT collaborators, Makoto 
Hirosawa, Masato Ishikawa and Masaki Hoshida. Hirosawa-san spent his whole winter 
vacation programming and running the folding algorithm. At the NIH, David Rawn 
(Towson State University) and I have made progress towards finding a topological prin-
ciple which unites the water seeking (hydrophilic) and water avoiding (hydrophobic) 
aspects of protein structure. A complete and simple topological model would reduce 
the N2 portion calculations to the number of amino acids in a given protein. With such 
a model we would hope to be able to fold proteins on many different types of computers. 

Discussion with the ICOT workers has also focused on computer languages, style 
of operating environment and network connectivity. Using the PSI II and III machines 
loaned to the NIH under the auspices of this collaboration, it has been possible to 
evaluate the state of development of the hardware and software produced by the Fifth 
Generation Project. Any user who decided to accept a research machine must know 
that it will be a lot of work. The FGCS conference shows that at the end of the project, 
much more of the potential of the hardware and software is now usable. During the 
continuation year we at the NIH would expect to make much greater use of the capa-
bilities of the PIM machines at ICOT. Discussions during the conference brought out 
- 57 -

the problems which other collaborators have been experiencing in the early utilization 
of the ICOT hardware and software. The decision by MITI announced at the confer-
ence is a clear indication that the Japanese viewpoints of the utility of international 
collaboration are rapidly changing. The PIMOS operating system should be ported to 
world-standard machines so that scientists all over the world can begin to do program 
development in KL1. 

During this collaborative project we have come to the opinion that even more than 
any single or parallel computer, the network is the most powerful artifact created by 
man. In this trans-global project we have experience the transition from paper letters 
to fax letters to fully electronic messages to interactive use of remote computers. In the 
beginning of the project the InterNet between Bethesda and Tokyo was rather slow and 
unreliable. Machines at either end had trouble talking to each other. As the project 
proceeded, the ability to communicate both electronically and intellectually rose to 
very high levels. The network gives us the ability to reason with each other about 
problems of mutual concern. The reasoning which we can do by sending messages is, 
however, rather limited. The InterNet is coming to the point where scientists can use 
databases and run processes on computers all over the world. New classes of tool for 
utilizing the network are being developed in many places in the world. We would hope 
to use these network tools to more strongly couple our collaborative research efforts. 

We thank the administrators of ICOT for making such a strong and exciting collab-
oration possible. We expect to continue our collaborative work long after the formal 
end of the project. 
- 58 -