{"id":"W4385431715","doi":"10.1016/j.meegid.2023.105484","title":"The utility of SARS-CoV-2 genomic data for informative clustering under different epidemiological scenarios and sampling","year":2023,"lang":"en","type":"article","venue":"Infection Genetics and Evolution","topic":"SARS-CoV-2 and COVID-19 Research","field":"Medicine","cited_by":3,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of British Columbia; BC Centre for Disease Control; Simon Fraser University","funders":"Michael Smith Health Research BC; Government of Canada; Australian Government","keywords":"Outbreak; Transmission (telecommunications); Biology; Cluster analysis; Genetic diversity; Population; Cluster (spacecraft); Sampling (signal processing); Evolutionary biology; Genetics; Statistics; Computer science; Medicine; Virology; Mathematics; Environmental health","routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":false},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007578051,0.00009139828,0.0001732714,0.00007882468,0.0002317663,0.00002510769,0.0000600989,0.0000883989,4.299468e-7],"category_scores_gemma":[0.0003330641,0.00006060386,0.00003524451,0.0001192896,0.0001440863,0.00005123109,0.0002710191,0.0001181121,0.000002156037],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00005219938,"about_ca_system_score_gemma":0.00004505189,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001515935,"about_ca_topic_score_gemma":0.0002033012,"domain_scores_codex":[0.9991795,0.00005119101,0.0002703667,0.000191231,0.0001076284,0.0002000971],"domain_scores_gemma":[0.9991933,0.0003646952,0.00007690684,0.000257726,0.00008856245,0.00001877778],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.0006048823,0.00007923731,0.7793881,0.0005106761,0.0001596595,4.797789e-7,0.0003519081,0.0001391975,0.1703947,0.0004274202,0.0006626836,0.04728102],"study_design_scores_gemma":[0.0004391577,0.0001584398,0.6459883,0.00002435518,0.00002568457,0.000008068616,0.0000774311,0.3445237,0.002430371,0.001541838,0.004731512,0.00005119117],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.937766,0.0004392765,0.06098582,0.0001743619,0.0001135863,0.0004162219,0.00002369353,0.00003071043,0.0000503236],"genre_scores_gemma":[0.9988244,0.0003337099,0.000164218,0.0005521587,0.0000718153,0.00002387429,0.00002111882,0.000006571953,0.000002115338],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.3443845,"threshold_uncertainty_score":0.2471354,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.2290169436114233,"score_gpt":0.4192692455366444,"score_spread":0.1902523019252211,"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."}}