The tRNAscan-SE interface
consists of three major components:
* Search mode selection
* Query sequence selection
* Configuration of search-mode and options for displaying results
The search mode determines which probabilistic model to use in searches – each model is based on tRNA training data from selected species or phylogenetic groups (for example, mammals, yeasts, archaea). If no explicit model for the species of interest is available in the user interface, specifying either a general model or a model from a related species generally yields good results. Different search modes can offer varying speed and sensitivity. For tRNAscan-SE, the standard search modes are fast and very sensitive in most instances. However, for short sequences that contain extremely atypical tRNAs, the “Infernal without HMM filter” mode offers slightly improved sensitivity.
Query sequence selection is used to specify the sequences to be searched for tRNAs. Raw or formatted sequence data can be pasted directly into the query sequence box or can be uploaded from a local file. Links to sample query sequence data are also available for demonstration purposes.
The server also has a set of program-specific search and output-display
options such as the tRNA isotype code. In addition, the server has an
adjustable cutoff score enabling tradeoffs between scan sensitivity and
specificity. In most cases, the default parameter choices will be
satisfactory and should be selected – especially by new users. However,
more experienced users are able to exert some control over the
program’s results by manipulating these parameters.
The output for tRNAscan-SE includes overall statistics as well as summary data for the entire search. The summary data includes counts of the total number of tRNAs found, the number of tRNA pseudogenes found, number of tRNAs with introns and which anticodons were detected. Finally, the output shows the predicted secondary structure for each identified sequence. Clicking on the “View” button of Predicted tRNA structure will provide a two-dimensional representation of the familiar tRNA clover-leaf structure. The output also displays the overall length of the sequence, the location of the anticodon and the overall tRNAscan-SE score.
Each predicted tRNA (not using mitochondrial tRNA search mode) is further evaluated with isotype-specific models. These 20+ models consist of the universal 20 amino acids, plus special models to discriminate initiator/formyl-methionine from elongator methionine, one to identify prokaryotic Ile tRNAs genomically encoded with a CAT anticodon, and one to recognize selenocysteine tRNAs. Alongside the predicted isotype based on the anticodon, the highest scoring isotype-specific model is also reported; any disagreement between the two functional prediction methods is reported for closer inspection. The scores of all isotype-specific models for each identified tRNA are shown for comparison.
The resulting tRNAs identified in the input sequence can be searched against the GtRNAdb by clicking on the “View” button of Similar tRNAs in GtRNAdb. Links to identical or highly similar tRNA genes found in the database will be listed. If a UCSC Genome Browser exists for any identical or close tRNA matches, a direct link is provided to those matches, enabling the user to examine its genomic context and any additional information available in genome browser tracks.
Chan, P.P., Lin, B., and Lowe, T.M. tRNAscan-SE 2.0. (In Preparation) Lowe, T.M. and Eddy, S.R. (1997) tRNAscan-SE: a program for improved detection of transfer RNA genes in genomic sequence. Nucleic Acids Research, 25, 955-964.