CZJ at CU.NIH.GOV
CZJ at CU.NIH.GOV
Mon Jul 22 19:40:20 EST 1991
I just returned from a week combining a US-EC taskforce on
biotechnology and a three day vacation in Amsterdam. Logging
onto my e-mail I found 248 messages from the various bboards.
Much of these messages concerned the desire to create versions of
AUTHORIN for additional platforms (UNIX and VMS) and the
possibility of enlisting 1000 points of light in creating these
First I would comment on the additional versions. GenBank is
funded primarily by the National Institutes of General Medical
Sciences under a contract awarded to Intelligenetics, Inc. with a
subcontract awarded to Los Alamos National Laboratories. The
work to be done is governed by the Statement of Work and the cost
determined by the proposal submitted in response to that
Statement of Work. The proposal submitted by IG did not propose
versions of AUTHORIN for four different platforms. Therefore,
any additional work, in this case new versions of AUTHORIN, will
require additional funds. The need for additional versions has
been discussed by the GenBank advisors. However, final
authorization can only come from the Contracting Officer,
following receipt of a formal proposal from IG and recommendation
from the Project Officers. Additional funds are not free. They
have to come from somewhere, i.e. someone's grant will go
unfunded. Hence the choices are often not easy, particularly if
your grant is the sacrificial lamb.
With respect to enlisting the scientific community in writing
code for these versions, the issue is not as simple as it first
appears. A lot of effort in commercial software development goes
into making a program easy to use with a simple, consistent,
intuitive interface. In addition the code must be written and
documented to certain standards to ensure easy maintenance.
Finally the output of AUTHORIN is a GenBank transaction which
must adhere to the protocol specified by LANL. This software
must be written to the highest professional standards. Just as
commercial software developers are not allowed mistakes which can
result in the lose of sales, AUTHORIN must turn users on, not
turn them off. The interface must work and be consistent and
intuitive. The output must be reliably parsable by the LANL
parser. Almost is not good enough. The software must undergo
extensive debugging and testing. None of this is glamorous but
essential. All of these processes are well understood by
commercial software developers but can represent boring hurdles
for the hacker. This is not to say that the proposal to let 1000
points of light shine is not possible, but I think the viability
could be difficult.
I hope these comments have been helpful.
One of the GenBank Project Officers.
More information about the Bioforum