tRNAscan-SE has been designed with the demands of human genome analysis in mind, but can be used for any DNA sequence. We estimate that tRNAscan-SE will detect about 99.5 % of the true tRNAs in the human genome, give zero false positives (except for tRNA-derived SINEs and tRNA pseudogenes), and take approximately 36 hours.
tRNAscan-SE demonstrates that general RNA structural profiles, covariance models, can be used as the basis for very sensitive RNA similarity searching. The primary limitation is speed. Although the strategy of using fast first-pass tRNA scanners in combination with second-stage covariance model analysis is effective here, this is not an attractive general strategy for searching for other RNA gene family members. Except for group I introns [Lisacek et al., 1994], there are no fast, specialized algorithms for detection of other RNA gene families, and much effort is required for creating these highly specialized new programs. Further work will focus on algorithmic development of covariance model search methods that will reduce both time and memory requirements, allowing faster searches for larger RNA genes without the need for first-pass screens.