{"id":{"repo_id":"houston","oai_identifier":"oai:uh-ir.tdl.org:10657/20842"},"canonical_url":"https://search.dev.ndltd.org/etd/houston/oai:uh-ir.tdl.org:10657/20842","repository":{"repo_id":"houston","name":"University of Houston","base_url":"https://uh-ir.tdl.org/server/oai/request"},"display":{"title":"Characterization of the Operational Trna Code for Amino Acids Through Supervised Machine Learning","abstract":"The operational tRNA code for amino acids is defined as a set of sequence- or structure-dependent rules governing the aminoacylation of tRNAs outside and independent of the anticodon. Previous efforts to uncover these rules have included experimental studies using synthetic and chemically modified transfer RNAs (tRNAs), as well as computational analyses aimed at predicting the cognate amino acid from tRNA primary sequences. In this study, we examined (1) whether tRNA primary sequences contain sufficient information to specify amino acid identity in the absence of the anticodon, (2) which regions are most important for aminoacylation, and (3) whether the operational code is universal or taxon-specific. Using supervised machine-learning classifiers trained on tRNA sequences, we found that the anticodon is not the sole determinant of tRNA identity. Even when the anticodon region was removed, classification accuracy remained high, indicating that other portions of the primary sequence encode meaningful recognition signals. Logistic-regression and odds-ratio analyses identified the acceptor stem and D-loop/stem as the regions most strongly associated with tRNA identity, whereas the anticodon and T-loop/stem were less consistently implicated. In contrast to the near universality of the classical genetic code, the operational RNA code for amino acids does not appear to be universal. Reciprocity tests revealed asymmetric predictability among species, suggesting species-specific features or potential model overfitting. Overall, our findings demonstrate that while tRNA recognition extends beyond the anticodon, the operational RNA code is not universal and is evolutionarily divergent. These results imply that tRNA identity is encoded through multiple, distributed sequence features that vary among species and amino acids, underscoring the importance of considering both structural and evolutionary contexts in future studies of tRNA–synthetase recognition.","abstract_html":"The operational tRNA code for amino acids is defined as a set of sequence- or structure-dependent rules governing the aminoacylation of tRNAs outside and independent of the anticodon. Previous efforts to uncover these rules have included experimental studies using synthetic and chemically modified transfer RNAs (tRNAs), as well as computational analyses aimed at predicting the cognate amino acid from tRNA primary sequences. In this study, we examined (1) whether tRNA primary sequences contain sufficient information to specify amino acid identity in the absence of the anticodon, (2) which regions are most important for aminoacylation, and (3) whether the operational code is universal or taxon-specific. Using supervised machine-learning classifiers trained on tRNA sequences, we found that the anticodon is not the sole determinant of tRNA identity. Even when the anticodon region was removed, classification accuracy remained high, indicating that other portions of the primary sequence encode meaningful recognition signals. Logistic-regression and odds-ratio analyses identified the acceptor stem and D-loop/stem as the regions most strongly associated with tRNA identity, whereas the anticodon and T-loop/stem were less consistently implicated. In contrast to the near universality of the classical genetic code, the operational RNA code for amino acids does not appear to be universal. Reciprocity tests revealed asymmetric predictability among species, suggesting species-specific features or potential model overfitting. Overall, our findings demonstrate that while tRNA recognition extends beyond the anticodon, the operational RNA code is not universal and is evolutionarily divergent. These results imply that tRNA identity is encoded through multiple, distributed sequence features that vary among species and amino acids, underscoring the importance of considering both structural and evolutionary contexts in future studies of tRNA–synthetase recognition.","abstract_has_math":false,"creators":["Gospodinov, Denis Ivov 1998-"],"institution":"University of Houston","degree_name":"Doctor of Philosophy","degree_level":null,"degree_discipline":"Biology","degree_department":null,"school":null,"contributors":[],"advisors":["Graur, Dan"],"committee_chairs":[],"committee_members":["Konstantinidis, Ioannis","Dauwalder , Brigitte","Daane, Jacob"],"year":2025,"date_issued":"2025-12","date_published":"2025-12","updated_at":"2026-07-24T02:32:54Z","subjects":["Transfer RNA","Aminoacyl tRNA synthetase","Supervised machine learning"],"languages":["English"],"rights":[],"rights_urls":[],"identifier_entries":[]},"links":{"outbound_url":"https://hdl.handle.net/10657/20842","outbound_label":"Handle","outbound_source":"dc:identifier.uri"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor.advisor","label":"Advisor","values":["Graur, Dan"]},{"key":"dc:contributor.committeemember","label":"Committee Member","values":["Konstantinidis, Ioannis","Dauwalder , Brigitte","Daane, Jacob"]},{"key":"dc:creator","label":"Author","values":["Gospodinov, Denis Ivov 1998-"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date.accessioned","label":"Dc Date Accessioned","values":["2026-02-05T16:48:54Z"]},{"key":"dc:date.issued","label":"Date","values":["2025-12"]},{"key":"dc:type","label":"Dc Type","values":["Thesis"]},{"key":"thesis:degree_discipline","label":"Discipline","values":["Biology"]},{"key":"thesis:degree_name","label":"Degree Name","values":["Doctor of Philosophy"]},{"key":"thesis:institution_name","label":"Thesis Institution Name","values":["University of Houston"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["Transfer RNA","Aminoacyl tRNA synthetase","Supervised machine learning"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language.iso","label":"Language (ISO)","values":["English"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier.uri","label":"Identifier URI","values":["https://hdl.handle.net/10657/20842"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description.abstract","label":"Abstract","values":["The operational tRNA code for amino acids is defined as a set of sequence- or structure-dependent rules governing the aminoacylation of tRNAs outside and independent of the anticodon. Previous efforts to uncover these rules have included experimental studies using synthetic and chemically modified transfer RNAs (tRNAs), as well as computational analyses aimed at predicting the cognate amino acid from tRNA primary sequences. In this study, we examined (1) whether tRNA primary sequences contain sufficient information to specify amino acid identity in the absence of the anticodon, (2) which regions are most important for aminoacylation, and (3) whether the operational code is universal or taxon-specific. Using supervised machine-learning classifiers trained on tRNA sequences, we found that the anticodon is not the sole determinant of tRNA identity. Even when the anticodon region was removed, classification accuracy remained high, indicating that other portions of the primary sequence encode meaningful recognition signals. Logistic-regression and odds-ratio analyses identified the acceptor stem and D-loop/stem as the regions most strongly associated with tRNA identity, whereas the anticodon and T-loop/stem were less consistently implicated. In contrast to the near universality of the classical genetic code, the operational RNA code for amino acids does not appear to be universal. Reciprocity tests revealed asymmetric predictability among species, suggesting species-specific features or potential model overfitting. Overall, our findings demonstrate that while tRNA recognition extends beyond the anticodon, the operational RNA code is not universal and is evolutionarily divergent. These results imply that tRNA identity is encoded through multiple, distributed sequence features that vary among species and amino acids, underscoring the importance of considering both structural and evolutionary contexts in future studies of tRNA–synthetase recognition."]},{"key":"dc:format.mimetype","label":"Dc Format Mimetype","values":["application/pdf"]},{"key":"dc:title","label":"Title","values":["Characterization of the Operational Trna Code for Amino Acids Through Supervised Machine Learning"]}]}],"canonical_facts":{"dc:contributor.advisor":["Graur, Dan"],"dc:contributor.committeemember":["Konstantinidis, Ioannis","Dauwalder , Brigitte","Daane, Jacob"],"dc:creator":["Gospodinov, Denis Ivov 1998-"],"dc:date.accessioned":["2026-02-05T16:48:54Z"],"dc:date.issued":["2025-12"],"dc:description.abstract":["The operational tRNA code for amino acids is defined as a set of sequence- or structure-dependent rules governing the aminoacylation of tRNAs outside and independent of the anticodon. Previous efforts to uncover these rules have included experimental studies using synthetic and chemically modified transfer RNAs (tRNAs), as well as computational analyses aimed at predicting the cognate amino acid from tRNA primary sequences. In this study, we examined (1) whether tRNA primary sequences contain sufficient information to specify amino acid identity in the absence of the anticodon, (2) which regions are most important for aminoacylation, and (3) whether the operational code is universal or taxon-specific. Using supervised machine-learning classifiers trained on tRNA sequences, we found that the anticodon is not the sole determinant of tRNA identity. Even when the anticodon region was removed, classification accuracy remained high, indicating that other portions of the primary sequence encode meaningful recognition signals. Logistic-regression and odds-ratio analyses identified the acceptor stem and D-loop/stem as the regions most strongly associated with tRNA identity, whereas the anticodon and T-loop/stem were less consistently implicated. In contrast to the near universality of the classical genetic code, the operational RNA code for amino acids does not appear to be universal. Reciprocity tests revealed asymmetric predictability among species, suggesting species-specific features or potential model overfitting. Overall, our findings demonstrate that while tRNA recognition extends beyond the anticodon, the operational RNA code is not universal and is evolutionarily divergent. These results imply that tRNA identity is encoded through multiple, distributed sequence features that vary among species and amino acids, underscoring the importance of considering both structural and evolutionary contexts in future studies of tRNA–synthetase recognition."],"dc:format.mimetype":["application/pdf"],"dc:identifier.uri":["https://hdl.handle.net/10657/20842"],"dc:language.iso":["English"],"dc:subject":["Transfer RNA","Aminoacyl tRNA synthetase","Supervised machine learning"],"dc:title":["Characterization of the Operational Trna Code for Amino Acids Through Supervised Machine Learning"],"dc:type":["Thesis"],"thesis:degree_discipline":["Biology"],"thesis:degree_name":["Doctor of Philosophy"],"thesis:institution_name":["University of Houston"]},"updated_at":"2026-07-24T02:32:54Z"}