This interesting methodological review--in three parts, with short coda--makes a strong case for integrating bioinformatics software.

Section 1.5 reviews a few of the existing “bioinformatics integration systems,” providing a rather complete table of comparison for eight of them, augmenting the text descriptions. If ever the need to develop an integrated system required an example, this section could easily serve as the poster child. All applications questions are purely abstract, and the reasons for failure are noted, but not detailed. That some scientists and some databases do not wish to access or be accessed is a critical issue, in need of exploration.

The authors move to “agent technology” in section 2, providing more conceptual models, delving into languages and their unique or common characteristics, and arguing for an agent communication language (ACL) and standards derived from it. These standards are considered important to the application proposed, but at times they are lost in a sea of undefined acronyms. Describing system characteristics, but lacking detailed examples, this section reads like a theory in search of an application.

The third section further expands the theme of “agents and integration,” arguing for wider application, in light of competing technologies. The fourth section, the coda, draws conclusions, including those inferred but not supported by data: the whole of data in biology and the biosciences has been, and will continue to be, a difficult area to integrate.

It will be interesting to observe over time if this model, when applied and implemented, will stand up well to the existing standards. Will it become the widely used reality? This is a readable document, broadly presented, on a highly technical issue. It could well serve as an outline for students interested in exploring the underlying question of integrating the diffusely held and highly variable knowledge content of the biology literature.