PROTEIN-COMPOUND OPTIMAL BINDING STRUCTURE PREDICTION METHOD USING LARGE-CAPACITY CONFORMER GENERATION AND THREE-DIMENSIONAL CONVOLUTIONAL DEEP TRANSFER LEARNING MODEL

The present invention relates to a method for predicting a protein-compound optimal binding structure (bestpose) using a three-dimensional convolutional deep learning model, which selects the optimal protein-compound binding structure (bestpose) by analyzing protein-compound binding structures using...

Full description

Saved in:
Bibliographic Details
Main Authors JUNG, Jong Sun, HONG, Jong Hui, KIM, Yong Hwan
Format Patent
LanguageEnglish
French
Korean
Published 28.03.2024
Subjects
Online AccessGet full text

Cover

Loading…
More Information
Summary:The present invention relates to a method for predicting a protein-compound optimal binding structure (bestpose) using a three-dimensional convolutional deep learning model, which selects the optimal protein-compound binding structure (bestpose) by analyzing protein-compound binding structures using a 3D-CNN model trained using a transfer learning method. The present invention is performed by comprising the steps of: (A) placing protein-compound binding sites in a three-dimensional space divided at preset intervals and calculating the density of atoms of proteins and compounds contained in each divided space region; and (B) setting, as a learning layer, binding information of the proteins and compounds included in each divided space, and learning the binding information by means of a convolutional neural network artificial intelligence algorithm so as to train a 3D-CNN model. According to the present invention, by training the 3D-CNN model using the transfer learning method, the binding environment of each protein with different compounds is reflected and learned, and thus, it is possible to provide a three-dimensional convolutional deep learning model having improved accuracy in protein-compound binding analysis. La présente invention concerne un procédé de prédiction d'une structure de liaison optimale de composé protéique (bestpose) à l'aide d'un modèle d'apprentissage profond convolutif tridimensionnel, qui sélectionne la structure de liaison de composé protéique optimale (bestpose) par analyse de structures de liaison de composé protéique à l'aide d'un modèle 3D-CNN entraîné à l'aide d'un procédé d'apprentissage de transfert. La présente invention est réalisée en comprenant les étapes consistant à : (A) placer des sites de liaison de composé protéique dans un espace tridimensionnel divisé à des intervalles prédéfinis et calculer la densité d'atomes de protéines et de composés contenus dans chaque région d'espace divisée ; et (B) définir, en tant que couche d'apprentissage, des informations de liaison des protéines et des composés inclus dans chaque espace divisé, et apprendre les informations de liaison au moyen d'un algorithme d'intelligence artificielle de réseau neuronal convolutif de façon à entraîner un modèle 3D-CNN. Selon la présente invention, par entraînement du modèle 3D-CNN à l'aide du procédé d'apprentissage par transfert, l'environnement de liaison de chaque protéine avec différents composés est réfléchi et appris, et ainsi, il est possible de fournir un modèle d'apprentissage profond convolutif tridimensionnel ayant une précision améliorée dans une analyse de liaison à un composé protéique. 본 발명은 전이학습(transfer learning)방법을 이용하여 학습된 3D-CNN모델을 이용하여 단백질-화합물 결합구조들을 분석하여 최적의 단백질-화합물 결합구조(bestpose)를 선별해내는 3차원 합성곱 심층 학습 모델을 이용한 단백질-화합물 최적결합구조 예측 방법에 관한 것으로, 본 발명은 (A) 단백질-화합물의 결합 부위를 기 설정된 간격으로 분할된 3차원 공간에 배치하고, 각 분할공간 영역에 포함된 단백질과 화합물의 원자의 밀도를 계산하는 단계와; (B) 각 분할공간의 포함된 단백질과 화합물의 결합정보를 학습레이어로 설정하여, 합성곱 신경망(Convolutional neural network) 인공지능 알고리즘으로 학습하여 3D-CNN 모델을 학습하는 단계;를 포함하여 수행된다. 이와 같은 본 발명에 의하면, 본 발명에서는 전이학습(transfer learning)방법을 이용한 3D-CNN모델을 학습시켜, 각 단백질의 서로 다른 화합물과의 결합 환경이 반영되어 학습되므로, 단백질-화합물 간 결합 분석의 정확도가 향상된 3차원 합성곱 심층 학습 모델을 제공할 수 있는 효과 있다.
Bibliography:Application Number: WO2023KR14451