{"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":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01338884,0.0008113041,0.0006375572,0.0024393,0.001029446,0.001840833,0.00110503,0.001385634,0.0007046941],"category_scores_gemma":[0.04677825,0.0004606729,0.001282543,0.001425039,0.001199986,0.001283763,0.001247717,0.001145362,0.0001775854],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001683961,"about_ca_system_score_gemma":0.001017551,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.006282777,"about_ca_topic_score_gemma":0.00542622,"domain_scores_codex":[0.9934652,0.00444167,0.0002933029,0.001190486,0.0004374887,0.0001717757],"domain_scores_gemma":[0.971885,0.02123201,0.001829326,0.002743801,0.0018354,0.0004744366],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"observational","study_design_scores_codex":[0.00139395,0.0003090247,0.1943437,0.0004380688,0.001013793,0.0004019912,0.001207376,0.7497159,0.01232837,0.007568587,0.001278693,0.03000062],"study_design_scores_gemma":[0.0001105546,0.0002224452,0.05540845,0.00008269543,0.0001675667,0.0002580857,0.0005086705,0.9231206,0.00556616,0.01328086,0.00116987,0.000104063],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9287991,0.0003374524,0.06675962,0.0004198305,0.00003079234,0.0001799741,0.002032757,0.0004057322,0.001034849],"genre_scores_gemma":[0.9596506,0.00005944426,0.03767734,0.00006540417,0.00001302595,0.00007327217,0.002318577,0.00005356194,0.00008873402],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.01338884,"threshold_uncertainty_score":0.07080781,"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."}}