STANFORD UNIVERSITY MEDICAL CENTER
STANFORD, CALIFORNIA 94305 ¢ (415) 321-1200

 

 

STANFORD UNIversiry SCHOOL oF MEDICINE December 16, 1970
Department of Genetics

Mr. Alan G, Janos
216 Bushnell.Hall
Kenyon College

P. O. Box 72
Gambier, Ohio 43022

Dear Mr. Janos,

Thank you for your letter of December llth. I am gratified at your
interest in the DENDRAL Program.

Unfortunately, the actual production of the documentation of our
system has not kept pace with our expectations or sometimes we have
used different routes than were implied in our earlier manuscripts.

Parts IV and V of the NASA report series have not appeared but
the entire package will be included in a book I am trying to put
together during the next year.

On the same account I have not completed the article originally
intended for "Accounts of Chemical Research".

I am sending you the Memo No, 62 that you requested plus some
additional writings.

You also request a listing of the source program, Please let me
explain why this requires further discussion. A program of this
complexity requires continuous attention for its proper maintenance.
The program is constantly being patched and revised as well, un-
fortunately, as having had to be moved from one computer system to
another several times during our experience with it. There is then
not one single definitive DENDRAL Program and in order to distribute
it we would need a clear mutual understanding of the expectations,
facilities, and competence of its recipient.

These problems are enormously magnified by the challenge of moving
DENDRAL on to any other and particularly a restricted computer system.
Our source program is written in the LISP language. I would be very
interested to know if there is any possibility of implementing LISP
on the 1130. I suspect not. However, if you do have a LISP capability

I will be glad to respond further about the adaptability of our source
text.

over

LT. J. P. KENNEDY, JR. LABORATORIES FOR MOLECULAR MEDICINE, DEDICATED TO RESEARCH IN MENTAL RETARDATION
MOLECULAR BIOLOGY HEREDITY NEUROBIOLOGY DEVELOPMENTAL MEDICINE
Mr. Alan G. Janos -~-2- 12/16/70

However, do not despair! It may well be both possible and
advantageous to re-program the core of the DENDRAL generator so
as to run it on a small machine. At worst this might even be
programmed in FORTRAN but you will then have many headaches in
organizing the dynamic allocation of storage for the assembly of
partial structures, With some ingenuity you may be able to trade
memory space for time in implementing the algorithm. In LISP-
DENDRAL it was very convenient to write the program recursively
but this does result in the accumulation of elaborate and voluminous
recursion stacks which you would have to avoid,

The essential core algorithm is summarized in Table 2 of the first
report, I am enclosing a copy marked to focus on the DENDRAL generator
for non-cyclic molecules. This is then documented in much greater detail
on pages 10 to 15 of Memo. No. 49 which is also on its way to you.

The various functions mentioned therein are all written in LISP and
will, I fear, almost certainly be jibberish to you. We would ordinarily
advocate that a colleague seriously interested in implementing DENDRAL
at his own installation spend a month or two working closely with us

to be able to work through the entire program. However, I believe you
should be able to visualize for yourself how each of the named functions
actually operates,

If you could go through the work of parsing through a particular
example, like all the isomers of acetic acid, using the general approach
outlined here I think you will be able to visualize what would have to
be implemented in a computer program.

This Memo describes an earlier version of DENDRAL which has since
been supercéeded be a table-driven model which gives much more flexibility
in dynamic re-arrangement of the priority of the different "values" listed
in Table 2. This is tidd in to the heuristics of applying DENDRAL to
solving problems of mass-spectrometry, some of which are amplified in great
detail in Memo No. 62.

As I indicated I appreciate your interest and that of Professor York
and hope that you will find some piece of this program a manageable challenge.
Please do not hesitate to call on me further —- particularly by phone or in
person if this is feasible ~ if I can be of further assistance,

Sincerely yours,

,
7 4 pawn Cus,

Joshua Lederberg
Professor of Genetics

JL/rr