Bacterial Diversity-Your Input Needed

Keith Robison robison1 at husc10.harvard.edu
Thu Jun 4 21:40:16 EST 1992


moorel at bionette.cgrb.orst.edu (Larry Moore) writes:

>I will be participating in a conference addressing the issue of
>establishing priorities in bacterial diversity research (15-18
>June).  In an effort to obtain a broad range of ideas, I'm
>requesting that anyone who has thoughts (including biases!) about
>this issue please e-mail them to me.

>It might be helpful to examine the following background statement
>relative to this request.
>----------------
>	A major limitation to identifying environmental isolates is
>	the absence of a well characterized and balanced database.
>	The current practice of using available databases which are
>	based primarily on medically important bacteria often does
>	not make it possible to identify environmental isolates.
>	   
	As a user of databases, I make the following comments.  
I hope they will help you.

>	   5.  How should these data be available to other
>	   scientists?  Should this be done through publication or
>	   databases or both?

Both.  Publish excerpts and summaries of the data, but maintain
electronic databases.  With any database showing much activity,
a published form is inherently out-of-date.  Only electronic
databases have the capability of being constantly up-to-date.
Also, it is cheaper to distribute things electronically than
physically.


>	   6.  What features are important to users of databases?

Consistency and accuracy.  It is better to have a small store of
well-kept data than a much larger mess of data.  To do this,
you _must_ commit to periodically reviewing all extant entries
and always review incoming entries.


>	   7.  How can we structure databases to reveal what are the
>	   most useful tests or comparisons that need to be done?

When you are designing your database structure, think in _broad_ terms.
The queries you use today may not be the ones you use tomorrow, so
make sure you don't design things too tightly around your current
questions.  Also, leave places to put information which doesn't fit
into your existing structure but _monitor_ those cubbyholes as they
will show you the weaknesses of your original scheme.

	If you do these things, you will find that users can rely
on your database and will like it.  If you don't, you will later
spend far more time and effort hunting down the problems and fixing
them.  Also, users will be unhappy and will not have confidence in
your work.  If you don't believe this, take a look at the nucleic acid
sequence databases.


>Larry Moore,
>moorel at cgrb.orst.edu


Keith Robison
Harvard University
Program in Biochemistry, Molecular, Cellular, and Developmental Biology

robison at ribo.harvard.edu 



More information about the Bioforum mailing list