A model for predicting drug-disease associations based on dense convolutional attention network

The development of new drugs is a time-consuming and labor-intensive process. Therefore, researchers use computational methods to explore other therapeutic effects of existing drugs, and drug-disease association prediction is an important branch of it. The existing drug-disease association predictio...

Descripción completa

Guardado en:
Detalles Bibliográficos
Autores principales: Huiqing Wang, Sen Zhao, Jing Zhao, Zhipeng Feng
Formato: article
Lenguaje:EN
Publicado: AIMS Press 2021
Materias:
Acceso en línea:https://doaj.org/article/029eeab8870949b59f8b66cf9f166132
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
id oai:doaj.org-article:029eeab8870949b59f8b66cf9f166132
record_format dspace
spelling oai:doaj.org-article:029eeab8870949b59f8b66cf9f1661322021-11-23T02:10:15ZA model for predicting drug-disease associations based on dense convolutional attention network10.3934/mbe.20213671551-0018https://doaj.org/article/029eeab8870949b59f8b66cf9f1661322021-08-01T00:00:00Zhttps://www.aimspress.com/article/doi/10.3934/mbe.2021367?viewType=HTMLhttps://doaj.org/toc/1551-0018The development of new drugs is a time-consuming and labor-intensive process. Therefore, researchers use computational methods to explore other therapeutic effects of existing drugs, and drug-disease association prediction is an important branch of it. The existing drug-disease association prediction method ignored the prior knowledge contained in the drug-disease association data, which provided a strong basis for the research. Moreover, the previous methods only paid attention to the high-level features in the network when extracting features, and directly fused or connected them in series, resulting in the loss of information. Therefore, we propose a novel deep learning model for drug-disease association prediction, called DCNN. The model introduces the Gaussian interaction profile kernel similarity for drugs and diseases, and combines them with the structural similarity of drugs and the semantic similarity of diseases to construct the feature space jointly. Then dense convolutional neural network (DenseCNN) is used to capture the feature information of drugs and diseases, and introduces a convolutional block attention module (CBAM) to weight features from the channel and space levels to achieve adaptive optimization of features. The ten-fold cross-validation results of the model DCNN and the experimental results of the case study show that it is superior to the existing drug-disease association predictors and effectively predicts the drug-disease associations.Huiqing WangSen ZhaoJing ZhaoZhipeng FengAIMS Pressarticledrug-disease association predictiongaussian interaction profile kernel similaritydense convolutional neural networkconvolutional block attention modulerandom forest classifierBiotechnologyTP248.13-248.65MathematicsQA1-939ENMathematical Biosciences and Engineering, Vol 18, Iss 6, Pp 7419-7439 (2021)
institution DOAJ
collection DOAJ
language EN
topic drug-disease association prediction
gaussian interaction profile kernel similarity
dense convolutional neural network
convolutional block attention module
random forest classifier
Biotechnology
TP248.13-248.65
Mathematics
QA1-939
spellingShingle drug-disease association prediction
gaussian interaction profile kernel similarity
dense convolutional neural network
convolutional block attention module
random forest classifier
Biotechnology
TP248.13-248.65
Mathematics
QA1-939
Huiqing Wang
Sen Zhao
Jing Zhao
Zhipeng Feng
A model for predicting drug-disease associations based on dense convolutional attention network
description The development of new drugs is a time-consuming and labor-intensive process. Therefore, researchers use computational methods to explore other therapeutic effects of existing drugs, and drug-disease association prediction is an important branch of it. The existing drug-disease association prediction method ignored the prior knowledge contained in the drug-disease association data, which provided a strong basis for the research. Moreover, the previous methods only paid attention to the high-level features in the network when extracting features, and directly fused or connected them in series, resulting in the loss of information. Therefore, we propose a novel deep learning model for drug-disease association prediction, called DCNN. The model introduces the Gaussian interaction profile kernel similarity for drugs and diseases, and combines them with the structural similarity of drugs and the semantic similarity of diseases to construct the feature space jointly. Then dense convolutional neural network (DenseCNN) is used to capture the feature information of drugs and diseases, and introduces a convolutional block attention module (CBAM) to weight features from the channel and space levels to achieve adaptive optimization of features. The ten-fold cross-validation results of the model DCNN and the experimental results of the case study show that it is superior to the existing drug-disease association predictors and effectively predicts the drug-disease associations.
format article
author Huiqing Wang
Sen Zhao
Jing Zhao
Zhipeng Feng
author_facet Huiqing Wang
Sen Zhao
Jing Zhao
Zhipeng Feng
author_sort Huiqing Wang
title A model for predicting drug-disease associations based on dense convolutional attention network
title_short A model for predicting drug-disease associations based on dense convolutional attention network
title_full A model for predicting drug-disease associations based on dense convolutional attention network
title_fullStr A model for predicting drug-disease associations based on dense convolutional attention network
title_full_unstemmed A model for predicting drug-disease associations based on dense convolutional attention network
title_sort model for predicting drug-disease associations based on dense convolutional attention network
publisher AIMS Press
publishDate 2021
url https://doaj.org/article/029eeab8870949b59f8b66cf9f166132
work_keys_str_mv AT huiqingwang amodelforpredictingdrugdiseaseassociationsbasedondenseconvolutionalattentionnetwork
AT senzhao amodelforpredictingdrugdiseaseassociationsbasedondenseconvolutionalattentionnetwork
AT jingzhao amodelforpredictingdrugdiseaseassociationsbasedondenseconvolutionalattentionnetwork
AT zhipengfeng amodelforpredictingdrugdiseaseassociationsbasedondenseconvolutionalattentionnetwork
AT huiqingwang modelforpredictingdrugdiseaseassociationsbasedondenseconvolutionalattentionnetwork
AT senzhao modelforpredictingdrugdiseaseassociationsbasedondenseconvolutionalattentionnetwork
AT jingzhao modelforpredictingdrugdiseaseassociationsbasedondenseconvolutionalattentionnetwork
AT zhipengfeng modelforpredictingdrugdiseaseassociationsbasedondenseconvolutionalattentionnetwork
_version_ 1718417357771112448