Adversarial Threshold Neural Computer for Molecular de Novo Design

In this article, we propose the deep neural network Adversarial Threshold Neural Computer (ATNC). The ATNC model is intended for the de novo design of novel small-molecule organic structures. The model is based on generative adversarial network architecture and reinforcement learning. ATNC uses a Di...

Celý popis

Uložené v:
Podrobná bibliografia
Vydané v:Molecular pharmaceutics Ročník 15; číslo 10; s. 4386
Hlavní autori: Putin, Evgeny, Asadulaev, Arip, Vanhaelen, Quentin, Ivanenkov, Yan, Aladinskaya, Anastasia V, Aliper, Alex, Zhavoronkov, Alex
Médium: Journal Article
Jazyk:English
Vydavateľské údaje: United States 01.10.2018
Predmet:
ISSN:1543-8392, 1543-8392
On-line prístup:Zistit podrobnosti o prístupe
Tagy: Pridať tag
Žiadne tagy, Buďte prvý, kto otaguje tento záznam!
Abstract In this article, we propose the deep neural network Adversarial Threshold Neural Computer (ATNC). The ATNC model is intended for the de novo design of novel small-molecule organic structures. The model is based on generative adversarial network architecture and reinforcement learning. ATNC uses a Differentiable Neural Computer as a generator and has a new specific block, called adversarial threshold (AT). AT acts as a filter between the agent (generator) and the environment (discriminator + objective reward functions). Furthermore, to generate more diverse molecules we introduce a new objective reward function named Internal Diversity Clustering (IDC). In this work, ATNC is tested and compared with the ORGANIC model. Both models were trained on the SMILES string representation of the molecules, using four objective functions (internal similarity, Muegge druglikeness filter, presence or absence of sp -rich fragments, and IDC). The SMILES representations of 15K druglike molecules from the ChemDiv collection were used as a training data set. For the different functions, ATNC outperforms ORGANIC. Combined with the IDC, ATNC generates 72% of valid and 77% of unique SMILES strings, while ORGANIC generates only 7% of valid and 86% of unique SMILES strings. For each set of molecules generated by ATNC and ORGANIC, we analyzed distributions of four molecular descriptors (number of atoms, molecular weight, logP, and tpsa) and calculated five chemical statistical features (internal diversity, number of unique heterocycles, number of clusters, number of singletons, and number of compounds that have not been passed through medicinal chemistry filters). Analysis of key molecular descriptors and chemical statistical features demonstrated that the molecules generated by ATNC elicited better druglikeness properties. We also performed in vitro validation of the molecules generated by ATNC; results indicated that ATNC is an effective method for producing hit compounds.
AbstractList In this article, we propose the deep neural network Adversarial Threshold Neural Computer (ATNC). The ATNC model is intended for the de novo design of novel small-molecule organic structures. The model is based on generative adversarial network architecture and reinforcement learning. ATNC uses a Differentiable Neural Computer as a generator and has a new specific block, called adversarial threshold (AT). AT acts as a filter between the agent (generator) and the environment (discriminator + objective reward functions). Furthermore, to generate more diverse molecules we introduce a new objective reward function named Internal Diversity Clustering (IDC). In this work, ATNC is tested and compared with the ORGANIC model. Both models were trained on the SMILES string representation of the molecules, using four objective functions (internal similarity, Muegge druglikeness filter, presence or absence of sp -rich fragments, and IDC). The SMILES representations of 15K druglike molecules from the ChemDiv collection were used as a training data set. For the different functions, ATNC outperforms ORGANIC. Combined with the IDC, ATNC generates 72% of valid and 77% of unique SMILES strings, while ORGANIC generates only 7% of valid and 86% of unique SMILES strings. For each set of molecules generated by ATNC and ORGANIC, we analyzed distributions of four molecular descriptors (number of atoms, molecular weight, logP, and tpsa) and calculated five chemical statistical features (internal diversity, number of unique heterocycles, number of clusters, number of singletons, and number of compounds that have not been passed through medicinal chemistry filters). Analysis of key molecular descriptors and chemical statistical features demonstrated that the molecules generated by ATNC elicited better druglikeness properties. We also performed in vitro validation of the molecules generated by ATNC; results indicated that ATNC is an effective method for producing hit compounds.
In this article, we propose the deep neural network Adversarial Threshold Neural Computer (ATNC). The ATNC model is intended for the de novo design of novel small-molecule organic structures. The model is based on generative adversarial network architecture and reinforcement learning. ATNC uses a Differentiable Neural Computer as a generator and has a new specific block, called adversarial threshold (AT). AT acts as a filter between the agent (generator) and the environment (discriminator + objective reward functions). Furthermore, to generate more diverse molecules we introduce a new objective reward function named Internal Diversity Clustering (IDC). In this work, ATNC is tested and compared with the ORGANIC model. Both models were trained on the SMILES string representation of the molecules, using four objective functions (internal similarity, Muegge druglikeness filter, presence or absence of sp3-rich fragments, and IDC). The SMILES representations of 15K druglike molecules from the ChemDiv collection were used as a training data set. For the different functions, ATNC outperforms ORGANIC. Combined with the IDC, ATNC generates 72% of valid and 77% of unique SMILES strings, while ORGANIC generates only 7% of valid and 86% of unique SMILES strings. For each set of molecules generated by ATNC and ORGANIC, we analyzed distributions of four molecular descriptors (number of atoms, molecular weight, logP, and tpsa) and calculated five chemical statistical features (internal diversity, number of unique heterocycles, number of clusters, number of singletons, and number of compounds that have not been passed through medicinal chemistry filters). Analysis of key molecular descriptors and chemical statistical features demonstrated that the molecules generated by ATNC elicited better druglikeness properties. We also performed in vitro validation of the molecules generated by ATNC; results indicated that ATNC is an effective method for producing hit compounds.In this article, we propose the deep neural network Adversarial Threshold Neural Computer (ATNC). The ATNC model is intended for the de novo design of novel small-molecule organic structures. The model is based on generative adversarial network architecture and reinforcement learning. ATNC uses a Differentiable Neural Computer as a generator and has a new specific block, called adversarial threshold (AT). AT acts as a filter between the agent (generator) and the environment (discriminator + objective reward functions). Furthermore, to generate more diverse molecules we introduce a new objective reward function named Internal Diversity Clustering (IDC). In this work, ATNC is tested and compared with the ORGANIC model. Both models were trained on the SMILES string representation of the molecules, using four objective functions (internal similarity, Muegge druglikeness filter, presence or absence of sp3-rich fragments, and IDC). The SMILES representations of 15K druglike molecules from the ChemDiv collection were used as a training data set. For the different functions, ATNC outperforms ORGANIC. Combined with the IDC, ATNC generates 72% of valid and 77% of unique SMILES strings, while ORGANIC generates only 7% of valid and 86% of unique SMILES strings. For each set of molecules generated by ATNC and ORGANIC, we analyzed distributions of four molecular descriptors (number of atoms, molecular weight, logP, and tpsa) and calculated five chemical statistical features (internal diversity, number of unique heterocycles, number of clusters, number of singletons, and number of compounds that have not been passed through medicinal chemistry filters). Analysis of key molecular descriptors and chemical statistical features demonstrated that the molecules generated by ATNC elicited better druglikeness properties. We also performed in vitro validation of the molecules generated by ATNC; results indicated that ATNC is an effective method for producing hit compounds.
Author Zhavoronkov, Alex
Putin, Evgeny
Ivanenkov, Yan
Aliper, Alex
Asadulaev, Arip
Vanhaelen, Quentin
Aladinskaya, Anastasia V
Author_xml – sequence: 1
  givenname: Evgeny
  surname: Putin
  fullname: Putin, Evgeny
  organization: Computer Technologies Lab , ITMO University , St. Petersburg 197101 , Russia
– sequence: 2
  givenname: Arip
  surname: Asadulaev
  fullname: Asadulaev, Arip
  organization: Computer Technologies Lab , ITMO University , St. Petersburg 197101 , Russia
– sequence: 3
  givenname: Quentin
  orcidid: 0000-0002-4611-2046
  surname: Vanhaelen
  fullname: Vanhaelen, Quentin
  organization: Pharma.AI Department , Insilico Medicine, Inc. , Baltimore , Maryland 21218 , United States
– sequence: 4
  givenname: Yan
  surname: Ivanenkov
  fullname: Ivanenkov, Yan
  organization: Institute of Biochemistry and Genetics Russian Academy of Science (IBG RAS) Ufa Scientific Centre , Oktyabrya Prospekt 71 , 450054 Ufa , Russian Federation
– sequence: 5
  givenname: Anastasia V
  surname: Aladinskaya
  fullname: Aladinskaya, Anastasia V
  organization: Moscow Institute of Physics and Technology (State University) , 9 Institutskiy Lane , Dolgoprudny City , Moscow Region 141700 , Russian Federation
– sequence: 6
  givenname: Alex
  surname: Aliper
  fullname: Aliper, Alex
  organization: Pharma.AI Department , Insilico Medicine, Inc. , Baltimore , Maryland 21218 , United States
– sequence: 7
  givenname: Alex
  orcidid: 0000-0001-7067-8966
  surname: Zhavoronkov
  fullname: Zhavoronkov, Alex
  organization: The Biogerontology Research Foundation , OX1 1RU Oxford , U.K
BackLink https://www.ncbi.nlm.nih.gov/pubmed/29569445$$D View this record in MEDLINE/PubMed
BookMark eNpNj8lOwzAYhC0Eogu8AjI3LileEic-lrJKpRzoPbKd3zTIiYMdV-LtqUSROM1o9M1IM0Onve8BoWtKFpQweqtMXHTeDTsVOmUgjYtSE0p5eYKmtMh5VnHJTv_5CZrF-EkIywvGz9GEyULIPC-m6G7Z7CFEFVrl8HYXIO68a_AGUjgEK98NaYSArQ_41TswyamAG8Abv_f4HmL70V-gM6tchMujztH748N29Zyt355eVst1prjgY1aABZILxhoiqCXGSC4JEcQ2pS6BAdWaKmNKbZUWh4qRklCrK1tVWpRsjm5-V4fgvxLEse7aaMA51YNPsWaEVoRKzuQBvTqiSXfQ1ENoOxW-67_X7Ac9GmBZ
CitedBy_id crossref_primary_10_1016_j_ymeth_2025_03_001
crossref_primary_10_1038_s42004_022_00733_0
crossref_primary_10_1016_j_compchemeng_2020_107005
crossref_primary_10_1002_jcc_26494
crossref_primary_10_1016_j_bioorg_2024_107285
crossref_primary_10_1111_cbdd_13674
crossref_primary_10_1186_s13321_025_00976_8
crossref_primary_10_1186_s13321_024_00887_0
crossref_primary_10_1016_j_medidd_2025_100213
crossref_primary_10_1002_advs_202101864
crossref_primary_10_1038_s41563_019_0338_z
crossref_primary_10_1039_D2SC05709C
crossref_primary_10_1002_cpt_1795
crossref_primary_10_1016_j_commatsci_2024_113258
crossref_primary_10_1002_minf_201900087
crossref_primary_10_1007_s12257_020_0049_y
crossref_primary_10_1080_17460441_2023_2134340
crossref_primary_10_1016_j_ailsci_2022_100031
crossref_primary_10_1080_14737140_2025_2557602
crossref_primary_10_1021_acs_jmedchem_4c01616
crossref_primary_10_1016_j_compbiolchem_2023_107911
crossref_primary_10_1186_s13321_022_00599_3
crossref_primary_10_1007_s00894_023_05492_w
crossref_primary_10_1021_acs_jcim_5c01020
crossref_primary_10_2174_1381612829666230428110542
crossref_primary_10_1016_j_drudis_2024_104133
crossref_primary_10_1016_j_ejphar_2025_177608
crossref_primary_10_1038_s41467_025_59870_4
crossref_primary_10_1002_slct_202502448
crossref_primary_10_1038_s42003_021_02586_0
crossref_primary_10_1073_pnas_1912707116
crossref_primary_10_1016_j_jmgm_2021_108045
crossref_primary_10_1002_wcms_1723
crossref_primary_10_1016_j_drudis_2023_103675
crossref_primary_10_1016_j_ces_2022_118119
crossref_primary_10_4155_fmc_2018_0358
crossref_primary_10_26599_BDMA_2023_9020009
crossref_primary_10_1016_j_ailsci_2022_100045
crossref_primary_10_1002_anie_201909987
crossref_primary_10_1007_s12039_023_02196_9
crossref_primary_10_1038_s41467_024_54456_y
crossref_primary_10_1016_j_ces_2025_122048
crossref_primary_10_1109_TNNLS_2021_3106392
crossref_primary_10_1038_s42256_024_00843_5
crossref_primary_10_1109_ACCESS_2023_3285811
crossref_primary_10_1146_annurev_biodatasci_092820_033938
crossref_primary_10_1186_s13321_019_0364_5
crossref_primary_10_1002_agt2_70089
crossref_primary_10_1002_jcc_27354
crossref_primary_10_1002_jcc_70100
crossref_primary_10_1039_D4SC00966E
crossref_primary_10_1039_D1RA03086H
crossref_primary_10_1080_17460441_2022_2019704
crossref_primary_10_1002_wcms_1395
crossref_primary_10_1109_TCBB_2024_3349990
crossref_primary_10_1021_acschemneuro_4c00405
crossref_primary_10_1016_j_ailsci_2023_100064
crossref_primary_10_1080_17460441_2024_2367014
crossref_primary_10_1002_ange_201909987
crossref_primary_10_3390_ijms22041676
crossref_primary_10_1021_acs_chemrev_4c00678
crossref_primary_10_1021_jacs_2c13467
crossref_primary_10_1039_D3SC04091G
crossref_primary_10_1016_j_ejmech_2024_117164
crossref_primary_10_1016_j_sbi_2021_10_001
crossref_primary_10_1021_acsomega_5c01443
crossref_primary_10_3390_ijms24076573
crossref_primary_10_1186_s13321_019_0397_9
crossref_primary_10_1007_s11427_023_2305_0
crossref_primary_10_1039_C9SC01928F
crossref_primary_10_1039_C9SC04026A
crossref_primary_10_1016_j_drudis_2024_103992
crossref_primary_10_4155_fmc_2021_0243
crossref_primary_10_1038_s42256_022_00463_x
crossref_primary_10_1038_s41598_019_47148_x
crossref_primary_10_1016_j_fmre_2024_11_027
crossref_primary_10_1002_wcms_1450
crossref_primary_10_3389_fphar_2020_565644
crossref_primary_10_1186_s13321_019_0328_9
crossref_primary_10_1038_s41587_019_0224_x
crossref_primary_10_1093_bib_bbab592
crossref_primary_10_1016_j_compbiomed_2022_105403
crossref_primary_10_1186_s13321_021_00497_0
crossref_primary_10_1038_s41467_021_23720_w
crossref_primary_10_1002_minf_202300183
crossref_primary_10_1038_s41598_020_79682_4
crossref_primary_10_3389_fmolb_2020_00136
crossref_primary_10_1146_annurev_chembioeng_101220_102232
crossref_primary_10_1039_D1SC01206A
crossref_primary_10_1016_j_neurot_2025_e00624
crossref_primary_10_1021_acs_jcim_5c01737
crossref_primary_10_1186_s13321_020_00473_0
crossref_primary_10_3390_ijms23052797
crossref_primary_10_1016_j_compbiolchem_2023_107927
crossref_primary_10_1038_s42256_021_00403_1
crossref_primary_10_3389_fgene_2024_1450529
crossref_primary_10_1016_j_compbiomed_2024_108810
crossref_primary_10_3390_molecules25143250
crossref_primary_10_1088_2632_2153_ad360e
crossref_primary_10_1002_jcc_27315
crossref_primary_10_1038_s42256_020_0209_y
crossref_primary_10_3389_fphar_2020_00269
crossref_primary_10_1038_s41587_020_0418_2
crossref_primary_10_6028_jres_126_008
crossref_primary_10_1039_C9ME00039A
crossref_primary_10_1186_s13321_021_00550_y
crossref_primary_10_1016_j_jbvi_2019_e00126
crossref_primary_10_3389_frhem_2024_1305741
ContentType Journal Article
DBID CGR
CUY
CVF
ECM
EIF
NPM
7X8
DOI 10.1021/acs.molpharmaceut.7b01137
DatabaseName Medline
MEDLINE
MEDLINE (Ovid)
MEDLINE
MEDLINE
PubMed
MEDLINE - Academic
DatabaseTitle MEDLINE
Medline Complete
MEDLINE with Full Text
PubMed
MEDLINE (Ovid)
MEDLINE - Academic
DatabaseTitleList MEDLINE
MEDLINE - Academic
Database_xml – sequence: 1
  dbid: NPM
  name: PubMed
  url: http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed
  sourceTypes: Index Database
– sequence: 2
  dbid: 7X8
  name: MEDLINE - Academic
  url: https://search.proquest.com/medline
  sourceTypes: Aggregation Database
DeliveryMethod no_fulltext_linktorsrc
Discipline Pharmacy, Therapeutics, & Pharmacology
EISSN 1543-8392
ExternalDocumentID 29569445
Genre Research Support, Non-U.S. Gov't
Journal Article
GroupedDBID ---
-~X
123
4.4
53G
55A
5VS
7~N
AABXI
ABJNI
ABMVS
ABQRX
ABUCX
ACGFS
ACS
ADHLV
AEESW
AENEX
AFEFF
AHGAQ
ALMA_UNASSIGNED_HOLDINGS
AQSVZ
BAANH
CGR
CS3
CUPRZ
CUY
CVF
DU5
EBS
ECM
ED~
EIF
EJD
F5P
GGK
GNL
H~9
IH9
JG~
NPM
P2P
RNS
ROL
UI2
VF5
VG9
W1F
7X8
ABBLG
ABLBI
ID FETCH-LOGICAL-a363t-5efe04622d061f0cc9390060fd7b7e2e1bb1acc7bfab6a36c9901fb8f88b672
IEDL.DBID 7X8
ISICitedReferencesCount 157
ISICitedReferencesURI http://www.webofscience.com/api/gateway?GWVersion=2&SrcApp=Summon&SrcAuth=ProQuest&DestLinkType=CitingArticles&DestApp=WOS_CPL&KeyUT=000446413400009&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D
ISSN 1543-8392
IngestDate Fri Jul 11 09:41:07 EDT 2025
Thu Apr 03 06:56:03 EDT 2025
IsPeerReviewed true
IsScholarly true
Issue 10
Keywords generative adversarial network
deep neural network
molecular de novo design
reinforcement learning
Language English
LinkModel DirectLink
MergedId FETCHMERGED-LOGICAL-a363t-5efe04622d061f0cc9390060fd7b7e2e1bb1acc7bfab6a36c9901fb8f88b672
Notes ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 23
ORCID 0000-0002-4611-2046
0000-0001-7067-8966
PMID 29569445
PQID 2018019329
PQPubID 23479
ParticipantIDs proquest_miscellaneous_2018019329
pubmed_primary_29569445
PublicationCentury 2000
PublicationDate 2018-10-01
20181001
PublicationDateYYYYMMDD 2018-10-01
PublicationDate_xml – month: 10
  year: 2018
  text: 2018-10-01
  day: 01
PublicationDecade 2010
PublicationPlace United States
PublicationPlace_xml – name: United States
PublicationTitle Molecular pharmaceutics
PublicationTitleAlternate Mol Pharm
PublicationYear 2018
SSID ssj0024523
Score 2.6140523
Snippet In this article, we propose the deep neural network Adversarial Threshold Neural Computer (ATNC). The ATNC model is intended for the de novo design of novel...
SourceID proquest
pubmed
SourceType Aggregation Database
Index Database
StartPage 4386
SubjectTerms Machine Learning
Neural Networks (Computer)
Title Adversarial Threshold Neural Computer for Molecular de Novo Design
URI https://www.ncbi.nlm.nih.gov/pubmed/29569445
https://www.proquest.com/docview/2018019329
Volume 15
WOSCitedRecordID wos000446413400009&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D
hasFullText
inHoldings 1
isFullTextHit
isPrint
link http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwpV05a8MwFBZtU0qX3kd6oUDJFKU-JWsqvUKXGEMzZDOSLEGH2mmdBvLv--QDT4VCF4ONBUI8vfe980PoNsx8asDsEUdyTQJKHSIc4ZEgNIGRIWUZMxXZBIvjaD7nSRNwK5uyylYnVoo6K5SNkYOT7oIyBbTB7xefxLJG2exqQ6GxiXo-QBlb0sXmUTdrL6zo3QAl-MQCgR00qIcVuHdCleOPwrazNoHjMZMg6j77HWlWFmey_9-9HqC9Bmvih1o4DtGGzo_QMKk3sx7hWdd7VY7wECfdGOv1MXqsuJpLYSUUfgW33GaqsJ3mAR9aNggMoBdPW45dnGkcF6sCP1eFISfobfIye3olDeMCET71lyTURttuVS8DM28cpbjP7cQWkzHJtKddKV2hFJNGSApLlM2qGRmZKJKUeadoKy9yfY6wMoDrOKAXX5sgYCYyRqrIpSrkVHMR9NGgPbkU5NkmKUSui-8y7c6uj87q408X9eCN1ANnjgdBePGH1Zdo177UdXdXqGfgNutrtK1Wy_fy66YSFHjGyfQHKFzKKw
linkProvider ProQuest
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=Adversarial+Threshold+Neural+Computer+for+Molecular+de+Novo+Design&rft.jtitle=Molecular+pharmaceutics&rft.au=Putin%2C+Evgeny&rft.au=Asadulaev%2C+Arip&rft.au=Vanhaelen%2C+Quentin&rft.au=Ivanenkov%2C+Yan&rft.date=2018-10-01&rft.eissn=1543-8392&rft.volume=15&rft.issue=10&rft.spage=4386&rft_id=info:doi/10.1021%2Facs.molpharmaceut.7b01137&rft_id=info%3Apmid%2F29569445&rft_id=info%3Apmid%2F29569445&rft.externalDocID=29569445
thumbnail_l http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/lc.gif&issn=1543-8392&client=summon
thumbnail_m http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/mc.gif&issn=1543-8392&client=summon
thumbnail_s http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/sc.gif&issn=1543-8392&client=summon