The Committee on Models for Biomedical Research has proposed a "matrix of biological knowledge" (Matrix) to address the need for a means to organize and intelligently access the wealth of information that confronts biomedical researchers (c.f. Models for Biomedical Research, National Academy Press, 1985). To implement the matrix will require a fusion of artificial intelligence (AI) and database technologies as well as an understanding of the structure of biological knowledge. However, a full implementation of the Matrix is best not attempted until pilot studies on a restricted domain of knowledge have demonstrated the feasibility and utility of the concept. One domain of considerable general interest is that of nucleic acid and protein sequence libraries, which have become invaluable research tools for a large segment of the biomedical research community. In addition to the libraries themselves a large number of computer programs have been written to analyze the sequence data, which has resulted in major new discoveries. The utility of these libraries to the typical laboratory investigator could be greatly enhanced by augmenting them with the types of knowledge management and reasoning systems envisioned for the Matrix. The goal of this proposed work is to build a testbed for the Matrix concept on top of the existing sequence libraries, benefitting both the larger goals of the Matrix and the immediate needs of sequence library users. Steps in this process will include: (1) implementation of the GenBank database in a relational form, interfaced to an AI environment; (2) creation of a series of knowledge bases relating to the database, starting with simple taxonomies and progression to more complex representations (on narrower test domains); (3) creation of a set of user interfaces oriented to the needs of molecular biologist; and (4) investigation of various techniques of inference against the knowledge base, in particular reasoning by analogy, with the ultimate aim of allowing users to build and test hypotheses locally.