Prediction and verification of mouse tRNAs identifies intron-containing families

Computer algorithms are often used to identify tRNA genes in newly sequenced genomes, but these predictions can be challenging. Not only are there structural variations and extremely limited sequence conservation among genes, but vertebrate genomes tend to have highly reiterated short interspersed sequences (SINEs) that originally derived from tRNA genes or tRNA-like transcription units. We have employed two programs, tRNAScan SE and ARAGORN, to predict the tRNA genes in the mouse nuclear genome, resulting in quite diverse but overlapping predicted gene sets. From these, we removed known SINE repeats and sorted the genes into predicted families and single-copy genes. In particular, four families of intron-containing tRNA genes were predicted, with introns in positions and structures analogous to the well characterized intron-containing tRNA genes in yeast. In this work we focus on verifying the expression of the intron-containing tRNA gene familes, as well as the other 30 tRNA gene familes. Keywords: tRNA, direct label Total RNa from 12 tissues was screened by microarray for predicted tRNAs.