{"id":"W4404976894","doi":"10.1186/s12859-024-05952-x","title":"Managing false positives during detection of pathogen sequences in shotgun metagenomics datasets","year":2024,"lang":"en","type":"article","venue":"BMC Bioinformatics","topic":"Salmonella and Campylobacter epidemiology","field":"Agricultural and Biological Sciences","cited_by":18,"is_retracted":false,"has_abstract":true,"ca_institutions":"Canadian Food Inspection Agency; Carleton University","funders":"Alliance de recherche numérique du Canada; Ontario Ministry of Agriculture, Food and Rural Affairs; Canadian Food Inspection Agency","keywords":"False positive paradox; Metagenomics; Shotgun; Computational biology; Shotgun sequencing; DNA microarray; Biology; True positive rate; Computer science; Bioinformatics; Genetics; DNA sequencing; Gene; Artificial intelligence; Gene expression","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":[0.03146903,0.002753855,0.001992199,0.003279349,0.002768088,0.003975935,0.003350751,0.003313374,0.0009750429],"category_scores_gemma":[0.08337077,0.001517737,0.003143159,0.00222477,0.00167862,0.003161981,0.002787989,0.003270067,0.0008897756],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001358006,"about_ca_system_score_gemma":0.002627103,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003286755,"about_ca_topic_score_gemma":0.008176018,"domain_scores_codex":[0.9833605,0.006004398,0.002116289,0.00481848,0.003021792,0.0006784995],"domain_scores_gemma":[0.9441187,0.04163821,0.003213179,0.005002435,0.005343195,0.0006843654],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.003949604,0.001063123,0.3333283,0.007159377,0.006660068,0.004081142,0.006157621,0.1253153,0.2262775,0.005746569,0.02294321,0.2573182],"study_design_scores_gemma":[0.0003195923,0.0007612729,0.06322713,0.0005919988,0.00154484,0.002159122,0.00136029,0.6753371,0.214052,0.02323745,0.01699307,0.0004162748],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.457819,0.003221004,0.5029445,0.001394346,0.0006352192,0.001085824,0.006779969,0.02431397,0.001806283],"genre_scores_gemma":[0.4467782,0.0003903206,0.5370497,0.001222379,0.000108787,0.0008534839,0.01113299,0.001953752,0.0005104659],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.03146903,"threshold_uncertainty_score":0.1664261,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.02605075273000819,"score_gpt":0.2442765274455539,"score_spread":0.2182257747155457,"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."}}