IDENTIFICATION OF NEW EUKARYOTIC TRANSFER-RNA GENES IN GENOMIC DNA DATABASES BY A MULTISTEP WEIGHT MATRIX ANALYSIS OF TRANSCRIPTIONAL CONTROL REGIONS

A linear method for the search of eukaryotic nuclear tRNA genes in DNA databases is described. Based on a modified version of the general we ight matrix procedure, our algorithm relies on the recognition of two intragenic control regions known as A and B boxes, a transcription ter mination signal, and on the evaluation of the spacing between these el ements. The scanning of the eukaryotic nuclear DNA database using this search algorithm correctly identified 933 of the 940 known tRNA genes (0.74% of false negatives). Thirty new potential tRNA genes were iden tified, and the transcriptional activity of two of them was directly v erified by in vitro transcription. The total false positive rate of th e algorithm was 0.014%. Structurally unusual tRNA genes, like those co ding for selenocysteine tRNAs, could also be recognized using a set of rules concerning their specific properties, and one human gene coding for such tRNA was identified. Some of the newly identified tRNA genes were found in rather uncommon genomic positions: 2 in centromeric reg ions and 3 within introns. Furthermore, the presence of extragenically located B boxes in tRNA genes from various organisms could be detecte d through a specific subroutine of the standard search program.