DELTA & Graphical User Interfaces

[maurobio at PUB2.LNCC.BR]

Posted on DELTA-L and TAXACOM

Dear DELTA Users,

Over the last two years or so, I've been deeply involved in
developing GUI-based tools for DELTA, of which my DIANA shells
(available for MS-DOS and Windows) were the first products.

I've already done *a lot* of reading on the issues of user
interface design, data models, search algorithms and data
structures and the relevance of all this to the development of
biological software. Perhaps more important, I've also undertaken
several experiments with user interfaces and data structures more
adequate to the presentation and representation (*sensu* Diederich
& Milton, 1993) of biological data. Here I intend to present some of
the preliminary, tentative conclusions I've reached in the course of
these experiments; I'm posting this to DELTA-L rather prematurely
because of the recent request of Andreas Walsh for advising on
GUI-development for DELTA. I hope to get some useful feedback from
DELTA users and developers.

Eventually, I can expand this "report" into an article to be
submitted to the DELTA Newsletter. Since Walsh's request
concerned only user interfaces, I'll not dwell here into the
issues of data representation (ie, database structures).

A key issue: portability
========================

I start from the principle that a Graphical User Interface is the
most desirable kind of computer environment for a computer user,
by making any system easier to grasp by the novice and faster to
use by the experienced user (but of course even this is open to
discussion... :-).

GUI's are available on the three main computational platforms in
use today - PC's, Macs, and UNIX workstations. The DELTA
programs, however, are limited to PC's. Following the several
discussions on this issue on the TAXACOM list (as well the *very
welcome* requests of some DIANA folks), I could verify that
perhaps more than offering a nice, intelligent, editing tool (more
on this below), the DELTA system should be *as portable as possible*,
not only because of the large number of both PC's and Macs users,
but also because of the widespread (and increasing) availability of
UNIX workstations. WWW browsers like Mosaic and Netscape offer a
relevant analogy here, since they are both available on the three
main platforms above mentioned.

So, one of my first conclusions is that an "ideal" DELTA system
should be as portable as possible.

However, developing a GUI-based program is not an easy task, even
for experienced programmers - the difficulties of writing such
programs are often underestimated. However, there are now
software tools that not only allow: a) to simplify the process of
software development for each of the widely used GUI platforms:
MS-Windows, MacIntosh, UNIX+XWindows, but also: b) to make the
programs portable among the mentioned platforms.

VIBRANT is one of these tools...

VIBRANT is a high-level, multi-platform user interface
development library written in C by Jonathan Kans
(kans at ray.nlm.nih.gov), Information Engineering Branch, National
Center for Biotechnology Information, NLM, NIH, Bethesda, USA. It
is distributed as part of the NCBI Software Development Toolkit.
VIBRANT acts as an intermediary between an application and the
underlying windowing system toolkit. It is available for free and
comes with full source codes. Using it, a GUI-based application can
be written that runs *without modification* on any of the mentioned
platforms - you just have to re-compile them on the target machine,
using the native C compiler.

Commercial software libraries based on the C language also exist
that perform similar functions - ZINC is one of the more
well-known of them - but these are usually very expensive and do
not include source code.

So, not only the DELTA programs themselves, but as a matter of
fact, all DELTA "third-party add-ons" (like DIANA, DMSWIN, and
TAXASOFT) should be greatly improved if (re-)written using such
tools, by benefiting from portability and, consequently, from a
larger user base.

A user-friendly editing tool
============================

Another old request from about all DELTA users is the
availability of a user-friendly, specialized data editor, that
should release users of the cumbersome and error-prone creation
and maintenance of DELTA datasets using an ordinary text editor
or word processor.

Early attempts to provide such a specialized DELTA editor were
those by Pankhurst's DEDIT and Gouda's DDATA (distributed as part
of his TAXASOFT package). However, none of these editors are
GUI-based, what may become a major drawback to a user
well-acquainted to electronic rodents... Worst, since MS-Windows
and other GUI environments have highly standard interfaces, with
common control objects (dialog boxes, pulldown menus, etc.), the
authors of these editors will have to write entirely new
applications when porting their programs to such an environment.

My specialized "DELTA Coding System" (DELCODE) is a first attempt
at providing a GUI-based highly interactive environment for the
editing of DELTA datasets. A DOS version is finished and a
Windows version is underway, but for several reasons I have not
yet released DELCODE to the public. Among the first of these is
that I do not consider it to be a "production tool", but rather a
teaching tool (that is the way I'm using it). It is the result of
one of the experiments I mentioned above, and is itself
experimental. DELCODE does *not* intend to be a rival of the
undoubtedly powerful, Windows-based, DELTA editor currently under
development by the DELTA development team. Its primary purpose is
to test algorithms, data structures, and user interface ideas. It
features pulldown menus, dialog boxes, and full mouse support,
but the DOS version is obviously not a full "GUI-based"
application (in the sense that it detaches at several points
from the CUA/SAA [Berry, 1988] standard).

Up to date, all DELTA editors implement what we could call a
"form fill-in" interface (see Diederich and Milton, 1993 - this
paper should be read by everyone wishing to develop GUI's for
biological software!), with which the user enters data, usually
from the keyboard, in specific fields on pre-built forms. This
system is also used by some of the Windows versions of the DELTA
programs (namely, INTKEY4 and INTIMATE), to get information from
the user. However, as one DIANA user has quite rightly pointed
out, character-by-taxon data matrices *are* the logical way to
manage taxonomic data. A matrix is convenient, already known to
most systematists, and is the accepted "metaphor" for comparative
data on taxa.

So, my second conclusion is that an "ideal" DELTA data entry and
management system should provide some sort of matrix-oriented
editor.

One such a matrix editor for taxonomic data already exists - and,
as far as I know, is not only *really wonderful* by itself, but
should provide a relevant paradigm for further development of
taxonomic data entry and management system. That is the editor of
Maddison & Maddison MacClade's interactive character analysis
program, available only for the MacIntosh computer. A good
summary description of this editor (and of MacClade itself) - for
those who do not have access to a Mac or do not have a copy of
MacClade - can be found in Maddison and Maddison (1989).

Another of my experiments with GUI software for DELTA was the
development of such a matrix-oriented editor for DELTA. A crude
prototype was implemented in Visual BASIC (that provides a rather
limited "grid" custom control), but I didn't develop it further.

Allowing the user to control her/his own TV...
==============================================

In recent years, the demands placed on software for management of
taxonomic data have increased dramatically in response to an
ever-increasing (and very welcome, by the way :-))) concern that
such data are the key to biodiversity conservation and
sustainable use of biological resources. As a result, the very
few software-developers engaged in the field of biodiversity
information management (including the members of the DELTA
development team) are being highly pressed to satisfy all user
needs in less time and with more efficiency. As other fields of
human activity, biodiversity data management is also facing the
'software crisis' (Cox, 1986; Gibbs, 1994), and perhaps software
engineering should also be enlisted as a 'crisis discipline',
along with conservation biology and cancer biology (Soule',
1985).

Such a crisis in the development of biological software means
that not only are better and portable software development tools
(as the VIBRANT interface library above mentioned, or the more
expensive commercial ones) needed, but also that a biodiversity
information management package far more *programmable* by the
end-user should be developed. The access to preprogrammed
subroutines would allow systematists with limited programming
experience to develop their own specialized programs for
biodiversity data management (eg., in her/his own national
language). In the future, it might be more fruitful for DELTA
developers to put their efforts in this general direction, rather
than trying to anticipate and meet (with scarce resources) the
precise needs of the next generation of computer taxonomists.
This is my third conclusion.

Along this line of thinking, I've developed DELTA Library, a
general-purpose library of routines for reading text files in
DELTA format, designed to work with any language that supports
Windows DLLs (Dynamic Link Libraries). They allow a programmer
using any language that supports DLL calling, such as Borland
Pascal/Turbo Pascal for Windows, C/C++, and Visual BASIC, to
write an application that can access DELTA datasets without
effort, with just one line of code for each routine's call.
I've been able to sucessfully integrate them with LORIS, an
experimental "DELTA viewer" I've written in Visual BASIC, and
these routines are also used by the Windows version of my DELCODE
editor for DELTA. While I had not yet released this library on a
widespread basis, it is free and available to anyone wishing to
try it.

I hope all this will start a fruitful discussion among the
subscribers of DELTA-L. I'm also in the process of setting up a
WWW personal home page, to facilitate the interchange of
information regarding this and related questions in biodiversity
data management and conservation. The above mentioned programs will
be put on our server and any interested user should be able to
download them through this page.

I look forward to hearing from you soon...

References
==========

Berry, R.E. 1988. Common User Access - a consistent and usable
   human-computer interface for the SAA environments. IBM Syst.
   J., 27: 281-300.

Cox, B.J. 1986. Object-Oriented Programming - An Evolutionary
   Approach. Addison-Wesley Publishing Company.

Diederich, J. and Milton, J. 1993. Expert workstations: a
   tools-based approach. In: Advances in Computer Methods for
   Systematic Biology: Artificial Intelligence, Databases,
   Computer Vision, R. Fortuner (Ed.). The Johns Hopkins
   University Press, Baltimore, Maryland, pp. 103-124.

Gibbs, W.W. 1994. Software's chronic crisis. Sci. Am., 271:
   72-81.

Maddison, W.P. and Maddison, D.R. 1989. Interactive analysis of
   phylogeny and character evolution using the computer program
   MacClade. Folia Primatol., 53: 190-202.

Soule', M.E. 1985. What is conservation biology? Bioscience, 35:
   727-734.

Warmest regards,

   + - - - - - - - - - - - Prof. Mauro J. Cavalcanti - - - - - - - - - - - +
   | Departamento de Biologia Geral              Universidade Santa Ursula |
   | Rua Fernando Ferrari, 75, Botafogo      22231-040, Rio de Janeiro, RJ |
   | Phone: +55 (021) 551-5542 ext 148                              BRASIL |
   |                     Internet: maurobio at omega.lncc.br                  |
   + - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - +