Title:
DEVICE AND METHOD FOR EXTRACTING COMPOUND INFORMATION
Document Type and Number:
WIPO Patent Application WO/2023/014007
Kind Code:
A1
Abstract:
The present disclosure provides a device and a method for extracting compound information. The method may comprise the steps of: processing input compound data in dimensions set for input to an encoder layer and a decoding layer; learning the input compound data by an attention method in the encoder layer; obtaining a mean vector and a variance vector that have a latent dimension, on the basis of the learned compound data in an information bottleneck layer; extracting a latent vector from a normal distribution according to the mean vector and the variance vector through re-parameterization; predicting physico-chemical properties of a compound on the basis of the mean vector in a compound property prediction layer; predicting the length of a compound sequence on the basis of the mean vector in a length prediction layer; converting the latent vector into encoder-output compound data having dimensions set for input to the decoding layer in an information extension layer; learning the input compound data by using the encoder-output compound data in an attention manner in the decoder layer; and reconstructing compound data from compound data learned in the decoder layer in a generator layer. The present disclosure may provide a compound information extraction model capable of extracting compound information that can be commonly used in compound prediction models for various purposes.
Inventors:
KIM DONGMIN (KR)
LEE MYUNGJAE (KR)
KANG SHINUK (KR)
HWANG HYUNJUN (KR)
KIM PYEONGEUN (KR)
LEE MYUNGJAE (KR)
KANG SHINUK (KR)
HWANG HYUNJUN (KR)
KIM PYEONGEUN (KR)
Application Number:
PCT/KR2022/011269
Publication Date:
February 09, 2023
Filing Date:
August 01, 2022
Export Citation:
Assignee:
JLK BIO INC (KR)
International Classes:
G16C20/30; G16C20/10; G16C20/40; G16C20/60; G16C20/70
Foreign References:
| KR20190089980A | 2019-07-31 |
Other References:
DOLLAR ORION, JOSHI NISARG, BECK DAVID A. C., PFAENDTNER JIM: "Attention-based generative models for de novo molecular design", CHEMICAL SCIENCE, ROYAL SOCIETY OF CHEMISTRY, UNITED KINGDOM, vol. 12, no. 24, 28 June 2021 (2021-06-28), United Kingdom , pages 8362 - 8372, XP093031436, ISSN: 2041-6520, DOI: 10.1039/D1SC01050F
ZHANG XIAO-CHEN, WU CHENG-KUN, YANG ZHI-JIANG, WU ZHEN-XING, YI JIA-CAI, HSIEH CHANG-YU, HOU TING-JUN, CAO DONG-SHENG: "MG-BERT: leveraging unsupervised atomic representation learning for molecular property prediction", BRIEFINGS IN BIOINFORMATICS, OXFORD UNIVERSITY PRESS, OXFORD., GB, vol. 22, no. 6, 5 November 2021 (2021-11-05), GB , pages bbab152 - bbab152-14, XP009543083, ISSN: 1467-5463, DOI: 10.1093/bib/bbab152
PRZEMYS{\L}AW SPUREK; TOMASZ DANEL; JACEK TABOR; MAREK \SMIEJA; {\L}UKASZ STRUSKI; AGNIESZKA S{\L}OWIK; {\L}UKASZ MAZIARKA: "Geometric Graph Convolutional Neural Networks", ARXIV.ORG, CORNELL UNIVERSITY LIBRARY, 201 OLIN LIBRARY CORNELL UNIVERSITY ITHACA, NY 14853, 11 September 2019 (2019-09-11), 201 Olin Library Cornell University Ithaca, NY 14853 , XP081481900
HYUNSEUNG KIM; JONGGEOL NA; WON BO LEE: "Generative chemical transformer: attention makes neural machine learn molecular geometric structures via text", ARXIV.ORG, CORNELL UNIVERSITY LIBRARY, 201 OLIN LIBRARY CORNELL UNIVERSITY ITHACA, NY 14853, 27 February 2021 (2021-02-27), 201 Olin Library Cornell University Ithaca, NY 14853 , XP081893618
ZHANG XIAO-CHEN, WU CHENG-KUN, YANG ZHI-JIANG, WU ZHEN-XING, YI JIA-CAI, HSIEH CHANG-YU, HOU TING-JUN, CAO DONG-SHENG: "MG-BERT: leveraging unsupervised atomic representation learning for molecular property prediction", BRIEFINGS IN BIOINFORMATICS, OXFORD UNIVERSITY PRESS, OXFORD., GB, vol. 22, no. 6, 5 November 2021 (2021-11-05), GB , pages bbab152 - bbab152-14, XP009543083, ISSN: 1467-5463, DOI: 10.1093/bib/bbab152
PRZEMYS{\L}AW SPUREK; TOMASZ DANEL; JACEK TABOR; MAREK \SMIEJA; {\L}UKASZ STRUSKI; AGNIESZKA S{\L}OWIK; {\L}UKASZ MAZIARKA: "Geometric Graph Convolutional Neural Networks", ARXIV.ORG, CORNELL UNIVERSITY LIBRARY, 201 OLIN LIBRARY CORNELL UNIVERSITY ITHACA, NY 14853, 11 September 2019 (2019-09-11), 201 Olin Library Cornell University Ithaca, NY 14853 , XP081481900
HYUNSEUNG KIM; JONGGEOL NA; WON BO LEE: "Generative chemical transformer: attention makes neural machine learn molecular geometric structures via text", ARXIV.ORG, CORNELL UNIVERSITY LIBRARY, 201 OLIN LIBRARY CORNELL UNIVERSITY ITHACA, NY 14853, 27 February 2021 (2021-02-27), 201 Olin Library Cornell University Ithaca, NY 14853 , XP081893618
Attorney, Agent or Firm:
PADO IP LAW PLLC (KR)
Download PDF:
Previous Patent: COMPOUND FOR TARGETED DEGRADATION OF RAS
Next Patent: FILTER MODULE AND AIR PURIFIER COMPRISING SAME
Next Patent: FILTER MODULE AND AIR PURIFIER COMPRISING SAME