tRNAscan-SE correctly predicted the introns for the 13 species of intron-containing tRNAs in the S. cerevisiae genome [Westaway & Abelson, 1995]. tRNAscan 1.3 often gives multiple intron predictions for each tRNA, making correct placement uncertain. EufindtRNA does not attempt to predict intron boundaries at all [Pavesi et al., 1994].
Detection of tRNAs containing long introns, usually group I or group II, is problematic. The default maximum tRNA length for tRNAscan-SE is 192 bp, but this can be increased (option -L <max length>) to allow searches with no practical limit on tRNA length. In the first phase of tRNAscan-SE, EufindtRNA searches for A and B boxes of the specified maximum distance apart, and passes only the 5' and 3' tRNA ends to covariance model analysis for confirmation (removing the bulk of long intervening sequences). Using this option, tRNAscan-SE was able to detect three of the four long tRNAs initially missed by all four methods in the Genbank tRNA subset search (the fourth tRNA was undetectable with EufindtRNA even with the intron removed before analysis). Group I or II introns in tRNAs tend to occur in positions other than the canonical position of protein-spliced introns, so tRNAscan-SE mispredicts the intron bounds and anticodon sequence for these cases. 5' and 3' tRNA bounds were correct for all three unusual tRNAs.