IUBio

NCBI needs help

frist at ccu.umanitoba.ca frist at ccu.umanitoba.ca
Fri Jul 3 11:41:54 EST 1992


Main points:

A. THE LINE BETWEEN PUBLIC AND PRIVATE IS SOMETIMES FUZZY
B. THERE IS A PLACE FOR PUBLICLY-SUPPORTED DATABASES AND SOFTWARE TOOLS 
C. DOES NCBI REALLY COMPETE WITH INDUSTRY?  
D. THERE IS ALSO A PLACE FOR COMMERCIAL SOFTWARE
E. SOFTWARE REPRESENTS A UNIQUE WAY TO SAVE MONEY IN RESEARCH.
F. WHERE DO WE DRAW THE LINE BETWEEN WHAT SHOULD BE DONE IN THE
   PRIVATE SECTOR OR THE PUBLIC SECTOR?

In article <Jul.2.19.51.44.1992.24544 at genbank.bio.net> kristoff at genbank.bio.net (David Kristofferson) writes:
>frist at ccu.umanitoba.ca writes:
>
>>Be careful- if you start producing your own freeware tools and (shudder)
>>give them away (you commies!), it might be considered economic terrorism!
>Let's "get real," folks.  None of the NCBI controversy extends to
>academic developers and the public domain software produced by them.

I disagree with you here. See B below.

A. THE LINE BETWEEN PUBLIC AND PRIVATE IS SOMETIMES FUZZY
 
>You may have forgotten this by now, but the first major public domain
>software source (the BIONET lending library) and all of these freely
>accessible newsgroups came from the BIONET National Computer Resource
>for Molecular Biology under a grant to a **commercial company.** If I
>remember correctly, you, Brian, were a donor to the lending library.
>The kind of exaggeration provided above simply obscures the issues.
>Let's look at some of these now.
I acknowledge your point as well taken. Academic researchers have benefited
from commercial concerns such as Intelligenetics, as per your example.
At the same time, Intelligenetics, while being a commercial firm, has
depended heavily on the public sector for support. It was founded using
the Stanford MOLGEN programs, and has also had a major influx of funds
from the GenBank contract. This is a case of the public sector paying
for something to be done, but having it done by a private contractor.
In a sense, the NCBI is only one step farther along in the same continuum.
They produce things for science with a budget appropriation instead of a 
contract. They employ people, and hire subcontractors (eg. to produce
conpact discs) and put money back into the economy in much the same
way that a private firm does. The main difference is that they don't
have to show a profit.

B. THERE IS A PLACE FOR PUBLICLY-SUPPORTED DATABASES AND SOFTWARE TOOLS 
>>In the case of NCBI, it's not just Entrez or Blast but the whole
>>developer's toolkit that they provide. In principle, this should save a 
>>great amount of reinvention of the wheel. The whole point here is that
>>you could NEVER get something like that from a commercial firm, because
>>it is not in their best interests to release source code. What they 
>>want to do is create more and more products. 
>>However, there is only  a very limited amount of money to pay for those
>>products. In my opinion, the availability of tools such as those from NCBI
>>is potentially a great boon to science, as well as a very economical
>>step. It costs the taxpayer a lot less if an agency like NCBI produces
>>tools that everyone can use. Shouldn't that be a consideration?
>
>I believe that Steve Smith at Harvard (and possibly others out there)
>have been working on user interfaces that are freely distributable and
>include source code.  That is perfectly within Steve's right to do so.

Here is one of my main points. There are many tasks that have not been
put into commercial packages, because commercial packages have to
identify the main things that everybody does and do these things
well. (There is some circularity here, because what people do is 
dependent on what's available to them). This means, though that
you can't plug new functionalities into commercial packages, a 
problem that has been very elegantly solved in Steve Smith's GDE.

This brings us to a more general issue, which is that much of the
academic software that appears in journals might as well not
exist, because of incompatibility between packages, and, more importantly,
the enormous amount of effort needed in porting to different systems
and in writing user interfaces (easily 90% of the work). Here,
NCBI has made a unique contribution: 

a) In the GenInfo backbone, they have standardized data representation
using the ASN.1 protocols, and at the same time provided a link between
some very diverse databases.
b) They have created a set of tools for interfacing with the backbone,
which saves the programmer from having to write routines to access the
database and for user interfaces. In the ideal, what this means is 
that someone like myself can concentrate on writing the code to do 
some new and useful thing with the database, and not have to waste a 
lot of time on the nuts and bolts, in particular with the user interface!

* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * 
The proposed limitations on NCBI DO interfere
with the academic software developers. We simply do not have the
resources to produce professional quality tools with slick user-interfaces
and inter-platform compatibility. Toolkits of the type that the
NCBI has developed are a way out of this dilemma. We will not
get them from industry, because it is not in their interest to
create them.
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * 

C. DOES NCBI REALLY COMPETE WITH INDUSTRY?  
>
>My understanding of *federal* regulations, however, is that the
>*government* is not allowed to compete against private industry.  NCBI
>and the commercial developers (which have been in this field for quite
>a few years) have tried to work out ways in which they can cooperate
>instead of compete, but this requires a careful delineation of
>responsibilities which I understand has been worked out at least in
>part.  Obviously NCBI plays an important role in database production
>and has also produced some very nice database searching software which
>we use on GOS (BLAST and IRX).  Just how broad of a mandate that they
>have, however, is one subject of controversy. 

I contend, for reasons given in a) and b), that NCBI is not competing
with private industry, which I why I was so mad at the suggestion that
they were. The creation of the tools and standards mentioned above
do not compete with private industry, because these are just not the
kind of things that industry can be expected to do. It would not be in their
best interests to create a toolkit that anyone can use, unless there
was a large enough market to be able to sell it. I would not be against
the use of this toolkit by industry to produce their own products.
If anything, that would add to the universality and intercomatibility 
of software for all concerned. What I do not want to happen is for
the standards and tools to be undermined by a group of companies who
would complain that they are losing a market that doesn't exist in
the first place.

D. THERE IS ALSO A PLACE FOR COMMERCIAL SOFTWARE
>As a hypothetical
>example, it would obviously be direct competition with the private
>sector if they were to distribute a comprehensive sequence analysis
>package because many of these have been available from a variety of
>sources for many years.

Yes, and on this point I am completely in agreement. The comercial
software products are very good, although many of them are 
outside the budgets of a lot of labs.  However, there is obviously
a market for them and I think that's fine. NCBI, as far as I know,
has not developed any sort of comprehensive seq. analysis package
that would compete with PC-GENE  or other products.
>

E. SOFTWARE REPRESENTS A UNIQUE WAY TO SAVE MONEY IN RESEARCH.
>
>Centrifuges and rotors cost scientists a **lot** of money, much more
>than a measly molbio software package,
Molbio software is not a measly cost. Several labs can share a centrifuge,
but each lab must but its own sequence software, or violate licensing
agreements. Few labs are going to buy more than one package, because
of the cost. A mini/mainframe system is even harder to set up, because
you typically have to have a paid systems person, and pay annual licensing
fees. At a great many institutions, it is simply impossible to convince
enough people that such a facility is necessary, and that they should
cough up the money. Here at the U. of M., our computing costs are
subsidized by the University. Consequently, I have been able to put 
together a sequence analysis/database facility with no budget, consisting
ENTIRELY OF FREEWARE. (See pub/psgendb/birch.ps.Z at ccu.umanitoba.ca)
 
> and we all know that resources
>are tight these days.  Why not, for "the good of science," create a
>federal agency that makes centrifuges and rotors and gives them away
>for free to all of the researchers in the country or charges at most a
>nominal fee???  This would save taxpayers money, right??  Why not do
>the same for DNA sequenators which cost in the $100,000 range???  Why
>should there be companies such as Beckman, Dupont, and ABI out there
>duplicating each other's efforts when we could attract all of the *best*
>instrument people to Washington and stop all of this waste and
>reinventing of the wheel???
>

I would contend that software is unique from other products and 
services. It costs a lot of money to build centrifuges and
rotors, to ship them and service them. Similarly, enzymes and chemicals
are consumable items that must constantly be produced. The costs of
production and shipping will be the same regardless of whether it
was done by a government agency or a private firm. 

In contrast, once a program is written, it can be distributed at 
minimal cost, particularly if available by FTP. In this w



More information about the Bio-soft mailing list

Send comments to us at biosci-help [At] net.bio.net