{"id":"W4327911455","doi":"10.1101/2023.03.16.532969","title":"Benchmarking of deep neural networks for predicting personal gene expression from DNA sequence highlights shortcomings","year":2023,"lang":"en","type":"preprint","venue":"bioRxiv (Cold Spring Harbor Laboratory)","topic":"Genomics and Chromatin Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":29,"is_retracted":false,"has_abstract":true,"ca_institutions":"Canadian Institute for Advanced Research","funders":"National Institutes of Health; Natural Sciences and Engineering Research Council of Canada; Canadian Institute for Advanced Research","keywords":"Benchmarking; Computational biology; DNA sequencing; Gene; Biology; Genome; Human genome; Genomics; Genetics; Deep learning; Artificial intelligence; Computer science; Machine learning","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":[0.005590277,0.001165154,0.0007865147,0.0009122355,0.0003628549,0.001276929,0.001928219,0.001279285,0.001961342],"category_scores_gemma":[0.01192313,0.0003690436,0.0006029345,0.001131955,0.0005695582,0.001499442,0.001056627,0.001510697,0.0009211971],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001197294,"about_ca_system_score_gemma":0.0009723697,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.01041701,"about_ca_topic_score_gemma":0.008519435,"domain_scores_codex":[0.9977581,0.0008448667,0.000152283,0.0005120326,0.0005678749,0.0001649474],"domain_scores_gemma":[0.9963256,0.001621727,0.0001083214,0.0008987595,0.0009070369,0.0001385793],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0003645501,0.0001653145,0.01192941,0.0001724636,0.0002314767,0.00007278406,0.00004862267,0.8485156,0.005262509,0.003628576,0.008880154,0.1207286],"study_design_scores_gemma":[0.00001114284,0.00004686961,0.001356349,0.00001873604,0.000009850588,0.00001668569,0.00001435773,0.9902059,0.004944605,0.002342822,0.001022574,0.00001012125],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.5744538,0.004429409,0.3835614,0.00302652,0.0007401361,0.0001865557,0.005607303,0.01343074,0.01456416],"genre_scores_gemma":[0.9055972,0.0005580607,0.08277784,0.0003766413,0.00005929915,0.0001553253,0.007670526,0.0004095905,0.002395665],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.01041701,"threshold_uncertainty_score":0.02956456,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.01399158335666094,"score_gpt":0.2189921514383153,"score_spread":0.2050005680816543,"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."}}