Interpretation of Compound Activity Predictions from Complex Machine Learning Models Using Local Approximations and Shapley Values
In qualitative or quantitative studies of structure–activity relationships (SARs), machine learning (ML) models are trained to recognize structural patterns that differentiate between active and inactive compounds. Understanding model decisions is challenging but of critical importance to guide comp...
Saved in:
| Published in | Journal of medicinal chemistry Vol. 63; no. 16; pp. 8761 - 8777 |
|---|---|
| Main Authors | , |
| Format | Journal Article |
| Language | English |
| Published |
United States
American Chemical Society
27.08.2020
|
| Subjects | |
| Online Access | Get full text |
| ISSN | 0022-2623 1520-4804 1520-4804 |
| DOI | 10.1021/acs.jmedchem.9b01101 |
Cover
| Abstract | In qualitative or quantitative studies of structure–activity relationships (SARs), machine learning (ML) models are trained to recognize structural patterns that differentiate between active and inactive compounds. Understanding model decisions is challenging but of critical importance to guide compound design. Moreover, the interpretation of ML results provides an additional level of model validation based on expert knowledge. A number of complex ML approaches, especially deep learning (DL) architectures, have distinctive black-box character. Herein, a locally interpretable explanatory method termed Shapley additive explanations (SHAP) is introduced for rationalizing activity predictions of any ML algorithm, regardless of its complexity. Models resulting from random forest (RF), nonlinear support vector machine (SVM), and deep neural network (DNN) learning are interpreted, and structural patterns determining the predicted probability of activity are identified and mapped onto test compounds. The results indicate that SHAP has high potential for rationalizing predictions of complex ML models. |
|---|---|
| AbstractList | In qualitative or quantitative studies of structure-activity relationships (SARs), machine learning (ML) models are trained to recognize structural patterns that differentiate between active and inactive compounds. Understanding model decisions is challenging but of critical importance to guide compound design. Moreover, the interpretation of ML results provides an additional level of model validation based on expert knowledge. A number of complex ML approaches, especially deep learning (DL) architectures, have distinctive black-box character. Herein, a locally interpretable explanatory method termed Shapley additive explanations (SHAP) is introduced for rationalizing activity predictions of any ML algorithm, regardless of its complexity. Models resulting from random forest (RF), nonlinear support vector machine (SVM), and deep neural network (DNN) learning are interpreted, and structural patterns determining the predicted probability of activity are identified and mapped onto test compounds. The results indicate that SHAP has high potential for rationalizing predictions of complex ML models.In qualitative or quantitative studies of structure-activity relationships (SARs), machine learning (ML) models are trained to recognize structural patterns that differentiate between active and inactive compounds. Understanding model decisions is challenging but of critical importance to guide compound design. Moreover, the interpretation of ML results provides an additional level of model validation based on expert knowledge. A number of complex ML approaches, especially deep learning (DL) architectures, have distinctive black-box character. Herein, a locally interpretable explanatory method termed Shapley additive explanations (SHAP) is introduced for rationalizing activity predictions of any ML algorithm, regardless of its complexity. Models resulting from random forest (RF), nonlinear support vector machine (SVM), and deep neural network (DNN) learning are interpreted, and structural patterns determining the predicted probability of activity are identified and mapped onto test compounds. The results indicate that SHAP has high potential for rationalizing predictions of complex ML models. In qualitative or quantitative studies of structure–activity relationships (SARs), machine learning (ML) models are trained to recognize structural patterns that differentiate between active and inactive compounds. Understanding model decisions is challenging but of critical importance to guide compound design. Moreover, the interpretation of ML results provides an additional level of model validation based on expert knowledge. A number of complex ML approaches, especially deep learning (DL) architectures, have distinctive black-box character. Herein, a locally interpretable explanatory method termed Shapley additive explanations (SHAP) is introduced for rationalizing activity predictions of any ML algorithm, regardless of its complexity. Models resulting from random forest (RF), nonlinear support vector machine (SVM), and deep neural network (DNN) learning are interpreted, and structural patterns determining the predicted probability of activity are identified and mapped onto test compounds. The results indicate that SHAP has high potential for rationalizing predictions of complex ML models. |
| Author | Rodríguez-Pérez, Raquel Bajorath, Jürgen |
| AuthorAffiliation | Department of Life Science Informatics, B-IT, LIMES Program Unit Chemical Biology and Medicinal Chemistry Department of Medicinal Chemistry Boehringer Ingelheim Pharma GmbH & Co. KG |
| AuthorAffiliation_xml | – name: Boehringer Ingelheim Pharma GmbH & Co. KG – name: Department of Medicinal Chemistry – name: Department of Life Science Informatics, B-IT, LIMES Program Unit Chemical Biology and Medicinal Chemistry |
| Author_xml | – sequence: 1 givenname: Raquel surname: Rodríguez-Pérez fullname: Rodríguez-Pérez, Raquel organization: Boehringer Ingelheim Pharma GmbH & Co. KG – sequence: 2 givenname: Jürgen orcidid: 0000-0002-0557-5714 surname: Bajorath fullname: Bajorath, Jürgen email: bajorath@bit.uni-bonn.de organization: Department of Life Science Informatics, B-IT, LIMES Program Unit Chemical Biology and Medicinal Chemistry |
| BackLink | https://www.ncbi.nlm.nih.gov/pubmed/31512867$$D View this record in MEDLINE/PubMed |
| BookMark | eNqFkUGP0zAQhS20iO0u_AOEfOTSMrYTN-FWVQus1BVILFwt25lQrxI72AnaXvnluJv2wgFOljXvex6_d0UufPBIyGsGKwacvdM2rR56bOwe-1VtgDFgz8iClRyWRQXFBVkAcL7kkotLcpXSAwAIxsULcilYyXgl1wvy-9aPGIeIox5d8DS0dBv6IUy-oRs7ul9uPNAvERtnj_NE2xj6J0mHj_RO273zSHeoo3f-B70LDXaJfkvHyy5Y3dHNMMTw6Ho98zobf93rjB_od91NmF6S563uEr46ndfk_sPN_fbTcvf54-12s1vqQsK4bE0BoinrSqA1-ZNo1saglaW0NTdQQlutJUDRCl0KNLpGKbVuWlMjN7IW1-TtbJvX-ZmfHVXvksWu0x7DlBTnVV2uhayP0jcn6WRywmqIef14UOfYsuD9LLAxpBSxVdbNAY5Ru04xUMeOVO5InTtSp44yXPwFn_3_g8GMPU3DFH0O69_IH0Q_rRA |
| CitedBy_id | crossref_primary_10_1002_int_23088 crossref_primary_10_1007_s11030_024_11061_x crossref_primary_10_1016_j_acra_2024_11_019 crossref_primary_10_1080_17470919_2023_2242094 crossref_primary_10_1039_D2DD00068G crossref_primary_10_4155_fmc_2023_0074 crossref_primary_10_1016_j_ajic_2021_11_012 crossref_primary_10_1016_j_scitotenv_2024_176575 crossref_primary_10_1016_j_leukres_2021_106639 crossref_primary_10_1016_j_hbpd_2022_11_005 crossref_primary_10_1063_5_0226151 crossref_primary_10_2214_AJR_24_31717 crossref_primary_10_1016_j_aiig_2023_09_002 crossref_primary_10_1021_acs_jcim_1c01163 crossref_primary_10_1002_biot_202400212 crossref_primary_10_1039_D5TC00305A crossref_primary_10_3892_ol_2024_14553 crossref_primary_10_1016_j_clnu_2021_11_006 crossref_primary_10_1007_s13748_022_00291_5 crossref_primary_10_1186_s12864_025_11434_y crossref_primary_10_1109_TCBB_2022_3190266 crossref_primary_10_1186_s40644_025_00855_3 crossref_primary_10_1039_D4MD00107A crossref_primary_10_1109_JSTARS_2022_3164140 crossref_primary_10_3389_fendo_2020_00643 crossref_primary_10_1371_journal_pone_0273831 crossref_primary_10_1016_j_ceramint_2022_06_156 crossref_primary_10_3390_ijms241915009 crossref_primary_10_1016_j_cpcardiol_2024_102456 crossref_primary_10_2340_1651_226X_2025_42982 crossref_primary_10_3389_fneur_2021_761092 crossref_primary_10_2196_67256 crossref_primary_10_3389_fnut_2023_1060398 crossref_primary_10_1002_pds_5756 crossref_primary_10_1016_j_crad_2023_10_011 crossref_primary_10_1016_j_ins_2023_119094 crossref_primary_10_1038_s41598_021_01099_4 crossref_primary_10_1016_j_ces_2024_119955 crossref_primary_10_1021_acs_jcim_2c00297 crossref_primary_10_1007_s11030_024_10809_9 crossref_primary_10_1186_s12880_023_01089_0 crossref_primary_10_3389_fcvm_2022_1042996 crossref_primary_10_1016_j_solener_2023_05_050 crossref_primary_10_3390_s24196203 crossref_primary_10_3390_toxics12110803 crossref_primary_10_1021_acs_jcim_3c00473 crossref_primary_10_3389_fped_2025_1519002 crossref_primary_10_4155_fmc_2022_0122 crossref_primary_10_1016_j_buildenv_2025_112875 crossref_primary_10_12677_acm_2025_152503 crossref_primary_10_1016_j_rineng_2025_104307 crossref_primary_10_3389_fbinf_2022_885983 crossref_primary_10_1016_j_compbiomed_2022_106379 crossref_primary_10_1002_minf_202200210 crossref_primary_10_1002_wcms_1568 crossref_primary_10_1002_env_2772 crossref_primary_10_1016_j_scitotenv_2022_159348 crossref_primary_10_1002_wcms_1681 crossref_primary_10_1016_j_ailsci_2024_100121 crossref_primary_10_1016_j_ailsci_2022_100048 crossref_primary_10_1007_s10994_023_06369_y crossref_primary_10_1007_s41870_024_02326_7 crossref_primary_10_1016_j_crad_2023_09_016 crossref_primary_10_1186_s13321_020_00479_8 crossref_primary_10_1016_j_xcrp_2023_101630 crossref_primary_10_4155_fdd_2021_0003 crossref_primary_10_1016_j_ijmedinf_2022_104930 crossref_primary_10_1021_acs_jcim_3c00465 crossref_primary_10_2147_JMDH_S448004 crossref_primary_10_1038_s41467_024_49979_3 crossref_primary_10_1016_j_xcrp_2024_102110 crossref_primary_10_1016_j_compbiolchem_2024_108324 crossref_primary_10_1016_j_isci_2022_105043 crossref_primary_10_1016_j_drudis_2020_12_003 crossref_primary_10_1021_acs_jcim_9b01188 crossref_primary_10_2215_CJN_0000000000000099 crossref_primary_10_1016_j_molliq_2024_126707 crossref_primary_10_1016_j_jconrel_2023_07_040 crossref_primary_10_1007_s11517_022_02569_1 crossref_primary_10_1177_17562848251321915 crossref_primary_10_1002_cam4_6540 crossref_primary_10_1088_2632_072X_ad1c68 crossref_primary_10_1016_j_ces_2025_121419 crossref_primary_10_1097_CORR_0000000000002276 crossref_primary_10_1186_s13321_022_00583_x crossref_primary_10_1002_cam4_4934 crossref_primary_10_1021_acs_jctc_2c01235 crossref_primary_10_3389_fmed_2022_1043412 crossref_primary_10_1186_s13244_024_01874_7 crossref_primary_10_1039_D4DD00257A crossref_primary_10_1007_s11096_025_01899_y crossref_primary_10_1007_s00330_021_08271_4 crossref_primary_10_1007_s10072_022_06351_x crossref_primary_10_1093_bib_bbad187 crossref_primary_10_1016_j_psychres_2021_114258 crossref_primary_10_1007_s11030_021_10239_x crossref_primary_10_1016_j_jmat_2024_07_006 crossref_primary_10_1186_s13321_021_00519_x crossref_primary_10_1002_adma_202415031 crossref_primary_10_1186_s13059_024_03225_7 crossref_primary_10_1038_s41598_024_71892_4 crossref_primary_10_1039_D4NP00009A crossref_primary_10_3934_DSFE_2024019 crossref_primary_10_4018_IJISP_303667 crossref_primary_10_1097_JS9_0000000000001082 crossref_primary_10_1021_jacs_1c05055 crossref_primary_10_1186_s12879_025_10450_3 crossref_primary_10_1007_s00521_024_10127_z crossref_primary_10_1038_s41598_022_10344_3 crossref_primary_10_1186_s12890_025_03480_9 crossref_primary_10_1007_s10822_020_00314_0 crossref_primary_10_1016_j_diii_2024_01_004 crossref_primary_10_1371_journal_pone_0303632 crossref_primary_10_1038_s42256_020_00236_4 crossref_primary_10_1245_s10434_024_16701_y crossref_primary_10_1007_s11606_023_08065_y crossref_primary_10_3389_fmed_2025_1521660 crossref_primary_10_3390_app12136681 crossref_primary_10_3389_fmedt_2025_1485244 crossref_primary_10_1016_j_diii_2024_01_008 crossref_primary_10_1016_j_cjche_2025_02_001 crossref_primary_10_1021_acs_jpcc_4c08108 crossref_primary_10_1111_ssqu_13459 crossref_primary_10_2139_ssrn_4178392 crossref_primary_10_3390_systems10040108 crossref_primary_10_1016_j_watres_2023_120470 crossref_primary_10_1080_17460441_2022_1985108 crossref_primary_10_1016_j_heliyon_2023_e22458 crossref_primary_10_3389_fdgth_2023_1193467 crossref_primary_10_1016_j_aquaeng_2023_102372 crossref_primary_10_1039_D2DD00084A crossref_primary_10_1021_acs_jcim_2c01091 crossref_primary_10_1038_s41598_023_45282_1 crossref_primary_10_1021_acs_jcim_0c01223 crossref_primary_10_1021_acs_jcim_0c01344 crossref_primary_10_1038_s41598_023_34650_6 crossref_primary_10_1021_acsomega_2c03812 crossref_primary_10_1016_j_ijbiomac_2024_136325 crossref_primary_10_1021_acsestwater_3c00290 crossref_primary_10_1371_journal_pone_0277257 crossref_primary_10_1016_j_ces_2023_118619 crossref_primary_10_1016_j_chempr_2024_02_004 crossref_primary_10_1016_j_toxlet_2024_02_012 crossref_primary_10_1002_minf_202200190 crossref_primary_10_1016_j_cub_2022_10_038 crossref_primary_10_1039_D2DD00136E crossref_primary_10_1186_s12885_024_12303_9 crossref_primary_10_1080_10643389_2024_2320753 crossref_primary_10_1038_s41467_024_49372_0 crossref_primary_10_1021_acs_jpca_4c06701 crossref_primary_10_1038_s41598_021_93771_y crossref_primary_10_1057_s41599_022_01106_7 crossref_primary_10_1016_j_foodchem_2023_138082 crossref_primary_10_3390_brainsci13040557 crossref_primary_10_1155_2022_6548800 crossref_primary_10_1007_s00432_024_05680_y crossref_primary_10_1007_s11426_024_2072_4 crossref_primary_10_1007_s10822_023_00528_y crossref_primary_10_1016_j_compenvurbsys_2024_102238 crossref_primary_10_1007_s10822_020_00361_7 crossref_primary_10_1039_D2RA05102H crossref_primary_10_1093_comnet_cnad030 crossref_primary_10_1016_j_tips_2023_04_002 crossref_primary_10_2196_35373 crossref_primary_10_1146_annurev_pharmtox_051921_023255 crossref_primary_10_1111_cbdd_14262 crossref_primary_10_1016_j_jmgm_2025_109016 crossref_primary_10_3389_fbioe_2021_662749 crossref_primary_10_3390_ijgi13100365 crossref_primary_10_1186_s13054_024_05245_y crossref_primary_10_3389_fpubh_2024_1453650 crossref_primary_10_3390_jpm12101739 crossref_primary_10_3390_app142311025 crossref_primary_10_1002_minf_202300022 crossref_primary_10_1007_s11063_023_11309_x crossref_primary_10_1016_j_ergon_2025_103699 crossref_primary_10_1016_j_eiar_2025_107895 crossref_primary_10_1021_acschemneuro_3c00840 crossref_primary_10_1016_j_clwat_2024_100041 crossref_primary_10_1016_j_compenvurbsys_2024_102133 crossref_primary_10_1016_j_patter_2022_100628 crossref_primary_10_1038_s42004_024_01393_y crossref_primary_10_1248_cpb_c23_00039 crossref_primary_10_1088_1741_2552_acf734 crossref_primary_10_1111_cas_16276 crossref_primary_10_1186_s12859_022_04760_5 crossref_primary_10_1016_j_acra_2024_08_024 crossref_primary_10_1016_j_scitotenv_2023_166623 crossref_primary_10_1016_j_acra_2024_10_039 crossref_primary_10_1080_17460441_2021_1918098 crossref_primary_10_1021_acs_jcim_2c01564 crossref_primary_10_1007_s11030_022_10555_w crossref_primary_10_3390_cryst12040556 crossref_primary_10_1016_j_xcrp_2021_100573 crossref_primary_10_1186_s13321_024_00820_5 crossref_primary_10_1002_cmdc_202400169 crossref_primary_10_1038_s41598_023_27858_z crossref_primary_10_1186_s13321_022_00673_w crossref_primary_10_4155_fmc_2020_0259 crossref_primary_10_1016_j_asoc_2023_110997 crossref_primary_10_2478_acss_2020_0016 crossref_primary_10_1016_j_ress_2025_110845 crossref_primary_10_1038_s41524_025_01552_2 crossref_primary_10_1016_j_dche_2024_100143 crossref_primary_10_3389_fonc_2022_848726 crossref_primary_10_1038_s41598_024_73733_w crossref_primary_10_1186_s12859_022_04663_5 crossref_primary_10_1016_j_rineng_2024_103071 crossref_primary_10_1016_j_jacc_2024_02_040 crossref_primary_10_3389_fcvm_2024_1454642 crossref_primary_10_1139_cjce_2023_0446 crossref_primary_10_1016_j_catena_2024_107848 crossref_primary_10_1007_s10143_021_01665_4 crossref_primary_10_1016_j_fuel_2025_134699 crossref_primary_10_2147_CLEP_S404580 crossref_primary_10_1016_j_indcrop_2024_118281 crossref_primary_10_1038_s41551_023_01034_0 crossref_primary_10_1124_pharmrev_122_000715 crossref_primary_10_1111_nyas_15113 crossref_primary_10_1021_acs_jcim_4c00126 crossref_primary_10_1021_acs_jcim_4c00366 crossref_primary_10_1007_s00330_022_08887_0 crossref_primary_10_1038_s41524_024_01282_x crossref_primary_10_1186_s13040_024_00378_w crossref_primary_10_1246_bcsj_20200220 crossref_primary_10_1371_journal_pone_0284315 crossref_primary_10_1039_D2RA07007C crossref_primary_10_1016_j_ailsci_2021_100009 crossref_primary_10_1021_acs_jcim_1c01263 crossref_primary_10_1021_acsomega_4c07994 crossref_primary_10_1177_14759217231184579 crossref_primary_10_1177_15330338231219352 crossref_primary_10_1080_10962247_2022_2035851 crossref_primary_10_1007_s11030_023_10649_z crossref_primary_10_1021_acs_jmedchem_1c01789 crossref_primary_10_1007_s10409_024_24155_x crossref_primary_10_3390_electronics14020272 crossref_primary_10_1039_D4EM00173G crossref_primary_10_1002_cmdc_202200434 crossref_primary_10_1016_j_acra_2024_11_037 |
| Cites_doi | 10.1021/jm4004285 10.1016/j.drudis.2014.02.004 10.1021/ci400737s 10.1038/nrd941 10.1021/acs.jcim.7b00088 10.1016/j.drudis.2014.10.012 10.1007/s10822-015-9860-5 10.1021/acs.jcim.5b00229 10.4155/fmc.11.23 10.1021/ci5005509 10.1021/acsomega.8b01682 10.1021/acsomega.9b00298 10.1021/jm901137j 10.1021/jm049228d 10.1021/acs.jmedchem.6b00906 10.1021/acsomega.7b01079 10.1080/10629360290002073 10.1080/17460441.2016.1201262 10.1039/C8SC00148K 10.1021/ci200528d 10.1007/s10822-007-9162-7 10.1021/jm00095a016 10.1093/nar/gkr777 10.1021/acs.jcim.8b00712 10.1016/0005-2795(75)90109-9 10.1021/acsomega.8b00462 10.1016/j.drudis.2018.05.010 10.1002/minf.201100136 10.1021/ci500410g 10.1186/s13321-014-0047-1 10.1021/acs.jcim.7b00146 10.1021/ci0601315 10.1007/s10822-008-9240-5 10.1145/2939672.2939778 10.1186/s13321-017-0232-0 10.1007/978-1-4757-3264-1 10.1038/s41551-018-0304-0 10.1007/BF01769885 10.1023/A:1010933404324 10.1007/978-0-387-84858-7 10.1002/minf.201100059 10.1021/acs.jcim.7b00087 10.1016/j.neunet.2005.07.009 10.1021/ci200409x 10.1021/acs.jcim.5b00175 10.1007/s11095-016-2029-7 10.1021/acs.jcim.5b00559 10.1021/acs.jcim.7b00274 10.1109/MSP.2012.2205597 10.1021/ci500747n 10.1109/ICPR.2010.764 10.1080/01621459.1973.10481460 10.1021/ci100050t |
| ContentType | Journal Article |
| DBID | AAYXX CITATION CGR CUY CVF ECM EIF NPM 7X8 |
| DOI | 10.1021/acs.jmedchem.9b01101 |
| DatabaseName | CrossRef Medline MEDLINE MEDLINE (Ovid) MEDLINE MEDLINE PubMed MEDLINE - Academic |
| DatabaseTitle | CrossRef MEDLINE Medline Complete MEDLINE with Full Text PubMed MEDLINE (Ovid) MEDLINE - Academic |
| DatabaseTitleList | MEDLINE - Academic MEDLINE |
| Database_xml | – sequence: 1 dbid: NPM name: PubMed url: https://proxy.k.utb.cz/login?url=http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed sourceTypes: Index Database – sequence: 2 dbid: EIF name: MEDLINE url: https://proxy.k.utb.cz/login?url=https://www.webofscience.com/wos/medline/basic-search sourceTypes: Index Database |
| DeliveryMethod | fulltext_linktorsrc |
| Discipline | Chemistry Pharmacy, Therapeutics, & Pharmacology |
| EISSN | 1520-4804 |
| EndPage | 8777 |
| ExternalDocumentID | 31512867 10_1021_acs_jmedchem_9b01101 f9777061 |
| Genre | Research Support, Non-U.S. Gov't Journal Article |
| GroupedDBID | - .K2 55A 5GY 5RE 5VS 7~N AABXI ABFLS ABMVS ABOCM ABPTK ABUCX ACGFS ACJ ACS AEESW AENEX AFEFF ALMA_UNASSIGNED_HOLDINGS AQSVZ BAANH CS3 DU5 EBS ED ED~ F5P GNL IH2 IH9 IHE JG JG~ K2 L7B LG6 P2P ROL TN5 UI2 VF5 VG9 W1F WH7 X XFK YZZ ZY4 --- -~X 4.4 6P2 AAHBH AAYXX ABBLG ABJNI ABLBI ABQRX ACGFO ADHLV AGXLV AHGAQ CITATION CUPRZ GGK XSW YQT ABTAH CGR CUY CVF ECM EIF NPM YIN 7X8 |
| ID | FETCH-LOGICAL-a460t-fb403d5983ecb520eb7bbec656c92b050f876004f3a53eba9e66aadfb9e2b693 |
| IEDL.DBID | ACS |
| ISSN | 0022-2623 1520-4804 |
| IngestDate | Fri Jul 11 01:00:56 EDT 2025 Wed Feb 19 02:30:32 EST 2025 Tue Jul 01 00:43:01 EDT 2025 Thu Apr 24 23:11:34 EDT 2025 Thu Oct 01 03:29:57 EDT 2020 |
| IsDoiOpenAccess | false |
| IsOpenAccess | true |
| IsPeerReviewed | true |
| IsScholarly | true |
| Issue | 16 |
| Language | English |
| License | http://pubs.acs.org/page/policy/authorchoice_termsofuse.html |
| LinkModel | DirectLink |
| MergedId | FETCHMERGED-LOGICAL-a460t-fb403d5983ecb520eb7bbec656c92b050f876004f3a53eba9e66aadfb9e2b693 |
| Notes | ObjectType-Article-1 SourceType-Scholarly Journals-1 ObjectType-Feature-2 content type line 23 |
| ORCID | 0000-0002-0557-5714 |
| OpenAccessLink | https://zenodo.org/record/3553762 |
| PMID | 31512867 |
| PQID | 2289573699 |
| PQPubID | 23479 |
| PageCount | 17 |
| ParticipantIDs | proquest_miscellaneous_2289573699 pubmed_primary_31512867 crossref_citationtrail_10_1021_acs_jmedchem_9b01101 crossref_primary_10_1021_acs_jmedchem_9b01101 acs_journals_10_1021_acs_jmedchem_9b01101 |
| ProviderPackageCode | JG~ 55A AABXI GNL VF5 7~N ACJ VG9 W1F ACS AEESW AFEFF .K2 ABMVS ABUCX IH9 BAANH AQSVZ ED~ UI2 CITATION AAYXX |
| PublicationCentury | 2000 |
| PublicationDate | 2020-08-27 |
| PublicationDateYYYYMMDD | 2020-08-27 |
| PublicationDate_xml | – month: 08 year: 2020 text: 2020-08-27 day: 27 |
| PublicationDecade | 2020 |
| PublicationPlace | United States |
| PublicationPlace_xml | – name: United States |
| PublicationTitle | Journal of medicinal chemistry |
| PublicationTitleAlternate | J. Med. Chem |
| PublicationYear | 2020 |
| Publisher | American Chemical Society |
| Publisher_xml | – name: American Chemical Society |
| References | ref9/cit9 ref45/cit45 ref3/cit3 ref27/cit27 ref63/cit63 ref16/cit16 Shapley L. S. (ref43/cit43) 1953 ref52/cit52 ref23/cit23 ref8/cit8 ref31/cit31 ref2/cit2 ref37/cit37 ref20/cit20 ref48/cit48 Krizhevsky A. (ref22/cit22) 2015; 25 ref10/cit10 Vapnik V. N. (ref17/cit17) 2000 ref53/cit53 ref19/cit19 ref42/cit42 Kenny P. W. (ref50/cit50) 2004 ref46/cit46 ref49/cit49 ref13/cit13 Lundberg S. M. (ref41/cit41) 2017 ref61/cit61 ref24/cit24 ref38/cit38 ref64/cit64 ref54/cit54 Nielsen M. A. (ref21/cit21) 2015 ref6/cit6 ref36/cit36 ref18/cit18 Bishop C. M. (ref60/cit60) 2006 ref65/cit65 Iooss B. (ref35/cit35) 2016 ref11/cit11 ref25/cit25 ref29/cit29 Boser B. E. (ref56/cit56) 1992 ref32/cit32 ref39/cit39 ref14/cit14 ref57/cit57 ref5/cit5 ref51/cit51 ref28/cit28 ref40/cit40 ref68/cit68 Hastie T. (ref34/cit34) 2009 ref26/cit26 ref55/cit55 ref69/cit69 ref12/cit12 ref15/cit15 ref62/cit62 Osborne M. J. (ref67/cit67) 1994 ref66/cit66 ref58/cit58 ref33/cit33 ref4/cit4 ref30/cit30 Pedregosa F. (ref59/cit59) 2011; 12 ref47/cit47 ref1/cit1 ref44/cit44 ref7/cit7 |
| References_xml | – ident: ref5/cit5 doi: 10.1021/jm4004285 – start-page: 307 volume-title: Contributions to the Theory of Games year: 1953 ident: ref43/cit43 – ident: ref19/cit19 doi: 10.1016/j.drudis.2014.02.004 – ident: ref6/cit6 doi: 10.1021/ci400737s – ident: ref8/cit8 doi: 10.1038/nrd941 – ident: ref58/cit58 doi: 10.1021/acs.jcim.7b00088 – ident: ref55/cit55 – ident: ref3/cit3 doi: 10.1016/j.drudis.2014.10.012 – ident: ref20/cit20 doi: 10.1007/s10822-015-9860-5 – ident: ref47/cit47 – ident: ref61/cit61 – ident: ref69/cit69 – ident: ref39/cit39 doi: 10.1021/acs.jcim.5b00229 – ident: ref40/cit40 doi: 10.4155/fmc.11.23 – ident: ref45/cit45 doi: 10.1021/ci5005509 – volume-title: A Course in Game Theory year: 1994 ident: ref67/cit67 – ident: ref31/cit31 doi: 10.1021/acsomega.8b01682 – ident: ref52/cit52 doi: 10.1021/acsomega.9b00298 – volume-title: Advances in Neural Information Processing Systems 30 year: 2017 ident: ref41/cit41 – ident: ref46/cit46 doi: 10.1021/jm901137j – ident: ref7/cit7 doi: 10.1021/jm049228d – start-page: 1 volume-title: Handbook of Uncertainty Quantification year: 2016 ident: ref35/cit35 – ident: ref49/cit49 doi: 10.1021/acs.jmedchem.6b00906 – ident: ref33/cit33 doi: 10.1021/acsomega.7b01079 – ident: ref37/cit37 doi: 10.1080/10629360290002073 – ident: ref24/cit24 doi: 10.1080/17460441.2016.1201262 – ident: ref62/cit62 – volume: 25 start-page: 1097 year: 2015 ident: ref22/cit22 publication-title: Adv. Neural Inf. Process. Syst. – ident: ref28/cit28 doi: 10.1039/C8SC00148K – ident: ref51/cit51 doi: 10.1021/ci200528d – ident: ref11/cit11 doi: 10.1007/s10822-007-9162-7 – ident: ref36/cit36 doi: 10.1021/jm00095a016 – ident: ref44/cit44 doi: 10.1093/nar/gkr777 – ident: ref12/cit12 doi: 10.1021/acs.jcim.8b00712 – ident: ref65/cit65 doi: 10.1016/0005-2795(75)90109-9 – ident: ref2/cit2 doi: 10.1021/acsomega.8b00462 – ident: ref4/cit4 doi: 10.1016/j.drudis.2018.05.010 – start-page: 271 volume-title: Cheminformatics in Drug Discovery year: 2004 ident: ref50/cit50 – ident: ref38/cit38 doi: 10.1002/minf.201100136 – ident: ref15/cit15 doi: 10.1021/ci500410g – ident: ref63/cit63 doi: 10.1186/s13321-014-0047-1 – ident: ref29/cit29 doi: 10.1021/acs.jcim.7b00146 – ident: ref9/cit9 doi: 10.1021/ci0601315 – ident: ref10/cit10 doi: 10.1007/s10822-008-9240-5 – ident: ref42/cit42 doi: 10.1145/2939672.2939778 – ident: ref54/cit54 – ident: ref27/cit27 doi: 10.1186/s13321-017-0232-0 – volume-title: The Nature of Statistical Learning Theory year: 2000 ident: ref17/cit17 doi: 10.1007/978-1-4757-3264-1 – ident: ref32/cit32 doi: 10.1038/s41551-018-0304-0 – ident: ref68/cit68 doi: 10.1007/BF01769885 – ident: ref16/cit16 doi: 10.1023/A:1010933404324 – volume-title: The Elements of Statistical Learning year: 2009 ident: ref34/cit34 doi: 10.1007/978-0-387-84858-7 – ident: ref14/cit14 doi: 10.1002/minf.201100059 – ident: ref30/cit30 doi: 10.1021/acs.jcim.7b00087 – ident: ref57/cit57 doi: 10.1016/j.neunet.2005.07.009 – ident: ref1/cit1 doi: 10.1021/ci200409x – ident: ref18/cit18 doi: 10.1021/acs.jcim.5b00175 – ident: ref25/cit25 doi: 10.1007/s11095-016-2029-7 – ident: ref48/cit48 doi: 10.1021/acs.jcim.5b00559 – ident: ref13/cit13 doi: 10.1021/acs.jcim.7b00274 – ident: ref23/cit23 doi: 10.1109/MSP.2012.2205597 – volume-title: Neural Networks and Deep Learning year: 2015 ident: ref21/cit21 – ident: ref26/cit26 doi: 10.1021/ci500747n – ident: ref64/cit64 doi: 10.1109/ICPR.2010.764 – start-page: 144 volume-title: Proceedings of the 5th Annual Workshop on Computational Learning Theory; Pittsburgh, Pennsylvania, 1992 year: 1992 ident: ref56/cit56 – volume: 12 start-page: 2825 year: 2011 ident: ref59/cit59 publication-title: J. Mach. Learn. Res. – ident: ref66/cit66 doi: 10.1080/01621459.1973.10481460 – volume-title: Pattern Recognition and Machine Learning year: 2006 ident: ref60/cit60 – ident: ref53/cit53 doi: 10.1021/ci100050t |
| SSID | ssj0003123 |
| Score | 2.7032137 |
| Snippet | In qualitative or quantitative studies of structure–activity relationships (SARs), machine learning (ML) models are trained to recognize structural patterns... In qualitative or quantitative studies of structure-activity relationships (SARs), machine learning (ML) models are trained to recognize structural patterns... |
| SourceID | proquest pubmed crossref acs |
| SourceType | Aggregation Database Index Database Enrichment Source Publisher |
| StartPage | 8761 |
| SubjectTerms | Deep Learning - statistics & numerical data Organic Chemicals - chemistry Support Vector Machine - statistics & numerical data |
| Title | Interpretation of Compound Activity Predictions from Complex Machine Learning Models Using Local Approximations and Shapley Values |
| URI | http://dx.doi.org/10.1021/acs.jmedchem.9b01101 https://www.ncbi.nlm.nih.gov/pubmed/31512867 https://www.proquest.com/docview/2289573699 |
| Volume | 63 |
| hasFullText | 1 |
| inHoldings | 1 |
| isFullTextHit | |
| isPrint | |
| link | http://utb.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwlV3fT9swELYmeBgvgzE2Ctt0kyYkJFISO3bqx6oaQtPYKlEm3iLbcdZBaRFpJcojf_nu8qMVmxDsNbIt2T77-5y7-46xz5n1MeK-CjpOqSB2kQhM4pNAmkhY6TMbacodPvmujs_ir-fyfPlQ_NuDz6ND44r2BQUrDv1VW1vCK3ztrHKFUENUqHe6uHlFxEWjDs4R15tUuUdGIUByxUNAeoRllmhztM5-NDk7VZDJZXs2tW1396-E4zMnssFe1cQTupWlvGYv_HiTvew19d422V6_UrGeH8BgmZRVHMAe9Jf61vM37P5hoCJMcqBbheozQddV1Sigf0MeoNKogVJYyiYjfwsnZfSmh1rY9RdQNbZRAWXsAnwjaIUuCZ3f_q6yKgswOPDp0GD3Ofw0I1y4LTY4-jLoHQd1KYfAxCqcBrmNQ5FJ3RHeWclDbxOL1oNk0mluQxnmHfIQxrkwUnhrtFfKmCy32nOrtHjLVsaTsd9mEGVIQnONTE0ksca2ymgrVRabKCe5-hbbx4VO65NYpKWTnUdp-bFe_bRe_RYTzdanrpZEp8ocoyd6BYte15UkyBPtPzVWleKWkkPGjP1kVqQcX7syEUrrFntXmdtiREFUrKOSnf-Yzy5b4_Q7IMTLL3nPVqY3M_8BOdPUfiwPyh_ivhXf |
| linkProvider | American Chemical Society |
| linkToHtml | http://utb.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwlV3da9swED9K99C97KP7yj5vMAqDOrMtS44eQ1jJtqQEmpW-GcmW121ZMuoEmj3uL9-dbCd0UEpfhXRIp5PupLv7HcC7wrqE9L4KerlSQZJHIjCpSwNpImGlK2ykOXd4fKyGX5PPZ_JsB2SbC0OTqIhS5Z34W3SB6AO3_eCYxXP3q6stqy169NyRKlFcsaE_ONlcwCKKRQsSHpN6bzPmrqHCeimvruqla4xNr3SO7sPpZro-1uRnd7W03fzPf0iOt17PA7jXmKHYr-XmIey4-T7sDdrqb_twMKkxrdeHON2maFWHeICTLdr1-hH8vRq2iIsS-Y7hak3Yz-vaFDi5YH-QF3HkhBbfZeYucexjOR02MK_fkGuzzSr0kQw4YkWLfYY9v_xe51hWaIjwybmh4Ws8NTPi32OYHn2cDoZBU9ghMLRby6C0SSgKqXvC5VbGobOpJVki0zLXsQ1lWPbYX5iUwkjhrNFOKWOK0moXW6XFE9idL-buGWBUkElaarLbRJpo6quMtlIViYlKBq_vwHtidNacyyrzLvc4ynxjw_2s4X4HRCsBWd4ApHOdjtkNo4LNqN81QMgN_d-2wpXRlrJ7xszdYlVlMb19ZSqU1h14WkvdhqJgw6yn0ue3WM8b2BtOx6Ns9On4ywu4G_NHQUjXYvoSdpcXK_eKrKmlfe3Pzj8FiB5B |
| linkToPdf | http://utb.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwnV3da9swEBelg20v--g-mm7rrjAKgzqzLEuOHkPW0PWLQLNR9mIkW263pUmpE2j2uL98d7KdtoNS1lchHZJ80p18d78fYx9y62K0-yroZEoFccZFYBKXBNJwYaXLLddUO3xwqHa-xrvH8vga1RdOokRJpQ_i06k-z4saYYB_ovaflLd46s7a2pLpwofPA_RJOLE2dHtHi0tY8Eg0QOERmvimau4WKWSbsvKmbbrF4fSGp_-UfV9M2eeb_GrPprad_f4HzfFea3rGntTuKHQr_XnOltx4hT3qNSxwK2xzUGFbz7dgeFWqVW7BJgyuUK_nL9ifm-mLMCmA7hpibYJuVnFUwOCC4kJe1YEKW3yXkbuEA5_T6aCGez0B4mgbleAzGmCfDC50Cf788kdVa1mCQcFHpwaHz-GbGeEevmTD_vawtxPUBA-BiVU4DQobhyKXuiNcZmUUOptY1Cl0MTMd2VCGRYfihnEhjBTOGu2UMiYvrHaRVVq8YsvjyditMuA5uqaFRv9NJLHGvspoK1UeG14QiH2LfcSNTuvzWaY-9B7x1DfWu5_Wu99iotGCNKuB0omvY3THqGAx6rwCCrmj_0ajYCl-UgrTmLGbzMo0wjewTITSusVeV5q3kCjIQeuoZO0_1vOePRx87qf7Xw733rDHEf0vCPF2TN6y5enFzL1Dp2pq1_3x-Qv29CC7 |
| openUrl | ctx_ver=Z39.88-2004&ctx_enc=info%3Aofi%2Fenc%3AUTF-8&rfr_id=info%3Asid%2Fsummon.serialssolutions.com&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Ajournal&rft.genre=article&rft.atitle=Interpretation+of+Compound+Activity+Predictions+from+Complex+Machine+Learning+Models+Using+Local+Approximations+and+Shapley+Values&rft.jtitle=Journal+of+medicinal+chemistry&rft.au=Rodr%C3%ADguez-P%C3%A9rez%2C+Raquel&rft.au=Bajorath%2C+J%C3%BCrgen&rft.date=2020-08-27&rft.eissn=1520-4804&rft.volume=63&rft.issue=16&rft.spage=8761&rft_id=info:doi/10.1021%2Facs.jmedchem.9b01101&rft_id=info%3Apmid%2F31512867&rft.externalDocID=31512867 |
| thumbnail_l | http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/lc.gif&issn=0022-2623&client=summon |
| thumbnail_m | http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/mc.gif&issn=0022-2623&client=summon |
| thumbnail_s | http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/sc.gif&issn=0022-2623&client=summon |