{"id":"W4406136011","doi":"10.1186/s12859-024-06030-y","title":"DFAST_QC: quality assessment and taxonomic identification tool for prokaryotic Genomes","year":2025,"lang":"en","type":"article","venue":"BMC Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":26,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"National Bioscience Database Center; Japan Society for the Promotion of Science; Ohsumi Frontier Science Foundation; Institute of Genetics; Japan Agency for Medical Research and Development","keywords":"Identification (biology); Computer science; Scalability; Genome; Biological classification; Data mining; Taxonomic rank; Data science; Biology; Database; Ecology","routes":{"ca_aff":false,"ca_fund":true,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":true},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0003382394,0.0001184525,0.0001418729,0.0000411803,0.0001287219,0.00005723038,0.0001104329,0.00007543641,0.000001210656],"category_scores_gemma":[0.00005178185,0.0001118419,0.00006236828,0.00004177199,0.00005823,0.000002175581,0.0001198741,0.00003063386,0.000002176046],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00002165529,"about_ca_system_score_gemma":0.0001336463,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000002455553,"about_ca_topic_score_gemma":0.00001535656,"domain_scores_codex":[0.9992264,0.00001828917,0.0003943086,0.0001562585,0.00004996281,0.0001547948],"domain_scores_gemma":[0.9994553,0.00003346431,0.0001336101,0.0002734662,0.00007811965,0.00002600456],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.0003954118,0.0003829363,0.4481682,0.004967391,0.001058604,2.096414e-7,0.000875522,0.001885557,0.3135827,0.04838892,0.01558618,0.1647083],"study_design_scores_gemma":[0.00251545,0.0003264839,0.8532566,0.00004075333,0.0001733327,0.000004782746,0.001365622,0.03570079,0.01996184,0.00268321,0.08324877,0.0007223496],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.7115092,0.001104824,0.2843259,0.0001414002,0.0002986022,0.001014851,0.0001039441,0.000007807118,0.001493437],"genre_scores_gemma":[0.8558947,0.0004112456,0.1420891,0.0002589362,0.00007399626,0.0001979199,0.0001200816,0.000009903211,0.0009441402],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.4050884,"threshold_uncertainty_score":0.4560782,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.03097355575466129,"score_gpt":0.3130307028387649,"score_spread":0.2820571470841037,"validation_status":"score_only:v0-immature-baseline","note":"Baseline scores from an immature model (maturity gate not passed). Scores rank; they never assert a category."}}