{"id":"W2566001817","doi":"10.1093/bioinformatics/btw793","title":"A new method for decontamination of <i>de novo</i> transcriptomes using a hierarchical clustering algorithm","year":2016,"lang":"en","type":"article","venue":"Bioinformatics","topic":"Genomics and Phylogenetic Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":52,"is_retracted":false,"has_abstract":true,"ca_institutions":"Université de Sherbrooke; Agriculture and Agri-Food Canada","funders":"Agriculture and Agri-Food Canada; U.S. Department of Agriculture","keywords":"Human decontamination; Computer science; Cluster analysis; Documentation; Sample (material); Scripting language; Identification (biology); Hierarchical clustering; Algorithm; Data mining; Biology; Artificial intelligence; Engineering; Programming language; Chemistry","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":[0.001504456,0.001701123,0.0009436583,0.002227798,0.001987232,0.00170668,0.002390467,0.001085912,0.004624928],"category_scores_gemma":[0.004214657,0.001024318,0.002269881,0.002181414,0.0008879449,0.00113946,0.001699802,0.00303124,0.003735],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001507913,"about_ca_system_score_gemma":0.002524512,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00884203,"about_ca_topic_score_gemma":0.01296755,"domain_scores_codex":[0.9978765,0.0002508998,0.0001816894,0.0006766056,0.0008856682,0.0001286475],"domain_scores_gemma":[0.9974244,0.0005729813,0.000252898,0.0006370388,0.0009683394,0.0001444359],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0003632161,0.0002068668,0.004948211,0.0009190085,0.0005777295,0.0003563685,0.0008520142,0.04035742,0.1983149,0.01301235,0.03470283,0.7053892],"study_design_scores_gemma":[0.0001316739,0.0001902859,0.007273582,0.0001195989,0.0002214118,0.0008854275,0.0002013386,0.7150676,0.167091,0.02032719,0.08823625,0.0002547285],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.004590999,0.0001007275,0.9873467,0.0001125905,0.0001225152,0.0001281939,0.0004465787,0.006457195,0.000694486],"genre_scores_gemma":[0.01261446,0.0000494124,0.9829717,0.00007666269,0.0000280125,0.000147767,0.001565782,0.001146721,0.001399467],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.00884203,"threshold_uncertainty_score":0.01758111,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.01805888345734189,"score_gpt":0.2906607890563882,"score_spread":0.2726019055990463,"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."}}