tRNAscan-SE 2.0 is an updated version of the widely used tool for identifying and annotating transfer RNA (tRNA) genes in genomes. The new version incorporates a larger training set, specialized isotype- and clade-specific models, and a comparative multi-model strategy to improve the detection and functional classification of tRNAs. Key enhancements include: 1. **Improved Detection and Classification**: The use of Infernal 1.1 covariance model search software and a suite of 76 different covariance models has significantly improved the sensitivity and specificity of tRNA gene detection. This allows for better identification and classification of both typical and atypical tRNAs. 2. **Specialized Models**: New and updated models are derived from a diverse set of over 4000 genomes, covering bacteria, archaea, and eukaryotes. These models are designed to capture the unique features of different tRNA isotypes, such as selenocysteine tRNAs and tRNAs with CAU anticodons. 3. **Comparative Analysis**: The program now employs a comparative multi-model strategy where predicted tRNAs are scored against a full set of isotype-specific covariance models. This allows for functional prediction based on both the anticodon and the highest-scoring isotype model, enhancing the accuracy of tRNA classification. 4. **Mitochondrial tRNA Detection**: tRNAscan-SE 2.0 includes a new mitochondrial tRNA search option, using clade- and isotype-specific covariance models trained on mitochondrial tRNA sequences. This enables accurate identification of mitochondrial tRNAs in vertebrate genomes. 5. **Intron Detection**: The program can now detect multiple and non-canonical introns in archaeal tRNA genes, improving the detection of atypical tRNA structures. 6. **High Confidence Filter**: A new 'high confidence' filter helps discriminate tRNA-derived repetitive elements from canonical tRNAs, providing more accurate annotations. Overall, tRNAscan-SE 2.0 maintains the high sensitivity and selectivity of the original program while offering significant improvements in accuracy and flexibility, making it a valuable tool for researchers in various fields.tRNAscan-SE 2.0 is an updated version of the widely used tool for identifying and annotating transfer RNA (tRNA) genes in genomes. The new version incorporates a larger training set, specialized isotype- and clade-specific models, and a comparative multi-model strategy to improve the detection and functional classification of tRNAs. Key enhancements include: 1. **Improved Detection and Classification**: The use of Infernal 1.1 covariance model search software and a suite of 76 different covariance models has significantly improved the sensitivity and specificity of tRNA gene detection. This allows for better identification and classification of both typical and atypical tRNAs. 2. **Specialized Models**: New and updated models are derived from a diverse set of over 4000 genomes, covering bacteria, archaea, and eukaryotes. These models are designed to capture the unique features of different tRNA isotypes, such as selenocysteine tRNAs and tRNAs with CAU anticodons. 3. **Comparative Analysis**: The program now employs a comparative multi-model strategy where predicted tRNAs are scored against a full set of isotype-specific covariance models. This allows for functional prediction based on both the anticodon and the highest-scoring isotype model, enhancing the accuracy of tRNA classification. 4. **Mitochondrial tRNA Detection**: tRNAscan-SE 2.0 includes a new mitochondrial tRNA search option, using clade- and isotype-specific covariance models trained on mitochondrial tRNA sequences. This enables accurate identification of mitochondrial tRNAs in vertebrate genomes. 5. **Intron Detection**: The program can now detect multiple and non-canonical introns in archaeal tRNA genes, improving the detection of atypical tRNA structures. 6. **High Confidence Filter**: A new 'high confidence' filter helps discriminate tRNA-derived repetitive elements from canonical tRNAs, providing more accurate annotations. Overall, tRNAscan-SE 2.0 maintains the high sensitivity and selectivity of the original program while offering significant improvements in accuracy and flexibility, making it a valuable tool for researchers in various fields.