In Silico Gene Selection for Custom Oligonucleotide Microarray Design

A method for systematically selecting the large number of sequences needed to custom design an oligonucleotide microarray was presented. This approach uses a Perl script to query sequence databases with gene lists obtained from previously designed (and publicly available) microarrays. Homologous sequences passing a user-defined threshold are returned and stored in a candidate gene database. Using this versatile technique, microarrays can be designed for any organism having sequence data. In addition, the ability to select specific input gene lists allows the design of microarrays tailored to address questions pertaining to specific pathways or processes. Given recent concerns about the accuracy of annotation in public sequence databases, it is also necessary to confirm the correct orientation of candidate sequences. This step is performed by a second Perl script that extracts protein similarity information from individual Unigene records, checks for consistency of features, and adds this information to the candidate gene database. Discrepancies between the orientations determined using protein similarities and that predicted by a given sequence’s assigned orientation are readily apparent by querying the candidate gene database.