In article <CMM.0.90.2.710263441.kristoff at genbank.bio.net> kristoff at GENBANK.BIO.NET (Dave Kristofferson) writes:
>>Tom,
>> I agree with virtually all of the points that you raise.
>Regarding BLAST, it is a very nice program and we use it on GOS (along
>with IRX which is another NLM production), but the recent introduction
>of BLAZE on GOS demonstrates that, contrary to the opinions of some,
>groups outside of NCBI can also make significant contributions in this
>area.
> Dave has conveniently forgotten a lot of history of commercial
companies in sequence analysis. With some very minor exceptions,
commercial companies, despite substantial capitalization in some cases,
have done virtually zero basic research in sequence analysis. At their
best, they have licensed software produced by academic developers and
repackaged it and supported it, but there are very few examples of their
having developed it.
Remember that NCBI is a relatively recent development (about 4
years old). There have been sequence analysis software companies for
almost ten years; one might ask, what did they develop during that
time? During the same time, there was very little government (or
private support) for sequence analysis software development - the
companies had the field to themselves, but did very very little. Even
today, very few companies are offering products that are based on the
state of the art 3 - 4 years ago. The argument that these companies
are being "held-back" by competition with government research is
nonsense; they didn't do research when they had no competition; BIONET
did very little research when it had a grant that required research as
well as user support (and thus the grant was not renewed); and there
is very little reason to think that they would do it today.
... material about interface design deleted ...
>Given the resources that NCBI has for research, it would also be
>unlikely that the small commercial concerns would be able to compete
>on the basis of new algorithm development.
We academic researchers, of course, must also compete with the
NCBI. I think we are doing so quite successfully, with research
budgets that are, in general, not only smaller than NCBI's but also
far smaller than those available to companies. I compete with NBCI,
and, although I suppose it is intimidating, I think it just means you
must be a bit more imaginative. I don't see the harm in that.
>A possible compromise, in
>my opinion, would be for NCBI to continue its database production and
>algorithm research roles, but to stay out of the user interface
>business. They might, of course, outline standards for database
>access interfaces, but general toolkit development seems to me, at
>least, to be overreaching their mandate.
The history of sequence analysis and sequence databases is
filled with examples of where attempts to commercialize databases and
software have backfired. Ten years ago, Genbank became popular, not
because DNA sequences were more informative than protein sequences but
because the Dayhoff protein sequence library was only available under
very restrictive and expensive terms. Likewise, molecular biology
algorithms that were licensed to commercial companies have disappeared
from general use - the Korn and Queen SEQ program and my restriction
mapping programs are examples with which I am familiar. The sequence
databases have been held back for years by infighting between the
different suppliers (EMBL, Genbank, PIR) so that one still cannot go
to a single sequence database and get cross references to the other
databases. This is another example of a relatively simple value-added
product that a company could have developed but did not; now NCBI is
doing it with the backbone database.
I think that the commercial software suppliers would have a
much stronger case if they could point to accomplishments that would
be stymied by competition with the NCBI. They can't. They can say,
"well, now there is no incentive" or "we can't compete with such a
large organization," but the fact it they weren't producing these
products when they had no competition, so why should we believe that
they would provide them now. Medline is an excellent example;
commercial distributors have done almost nothing to improve the user
interface to Medline; it has taken NCBI and ENTREZ to revolutionize
access to that database. The commercial distributors had more than 10
years to introduce novel ways to access Medline; instead they simply
provided slightly modified versions of the original National Library
of Medicine system.
The attempt to stop the NCBI is another example of companies
spending their research budgets on lawyers and lobbyists rather than on
research.
Bill Pearson
