| Over the past 15 years, bioinformatics has played an increasingly important role in pharmaceutical research. This dissertation presents two bioinformatics applications that may have particular relevance to drug discovery or development. The first application (Chapter 2) describes the development of two new algorithms (called GRPL and GRPL+) to improve the accuracy of gene prediction in eukaryotic DNA. Specifically, this set of computer program combines reference point logistic (RPL) methods with “smart” sequence alignment methods to provide substantially improved accuracy in gene identification. As part of the evaluation of the GRPL program, we show how the quality of gene predictions can be improved with increasing database size. This technique has important implications for identifying known and unknown disease genes in the draft human genome sequence. Chapter 3 describes the development of self-updating, self-correcting databases containing biological or chemical information. Self-updating databases use data-mining and data-validation methods to automatically extract and deposit “corrected” electronic information into a database or archive. With the rapid increase in both the size and number of biological, chemical and pharmaceutical databases, there is a growing need to create automated methods to consolidate, update, validate and correct the data in these databases. In this thesis, I illustrate the development of such a self-updating, self-correcting database with a specific example called RefDB. RefDB is a database used by NMR spectroscopists to archive, access and analyze NMR chemical shift data. |
