Pub Date : 2023-09-05DOI: 10.2174/1574893618666230905102407
Adeel Malik, Jamal S. M. Sabir, M. Kamli, Thi Phan Le, Chang-Bae Kim, Balachandran Manavalan
Krüppel-like factors (KLFs) are a family of transcription factors containing zinc fingers that regulate various cellular processes. KLF proteins are associated with human diseases, such as cancer, cardiovascular diseases, and metabolic disorders. The KLF family consists of 18 members with diverse expression profiles across numerous tissues. Accurate identification and annotation of KLF proteins is crucial, given their involvement in important biological functions. Although experimental approaches can identify KLF proteins precisely, large-scale identification is complicated, slow, and expensive. In this study, we developed RDR100, a novel random forest (RF)-based framework for predicting KLF proteins based on their primary sequences. First, we identified the optimal encodings for ten different features using a recursive feature elimination approach, and then trained their respective model using five distinct machine learning (ML) classifiers. Results: The performance of all models was assessed using independent datasets, and RDR100 was selected as the final model based on its consistent performance in cross-validation and independent evaluation. Our results demonstrate that RDR100 is a robust predictor of KLF proteins. RDR100 web server is available at https://procarb.org/RDR100/.
{"title":"RDR100: an effective computational method for identifying Kruppel-like factors","authors":"Adeel Malik, Jamal S. M. Sabir, M. Kamli, Thi Phan Le, Chang-Bae Kim, Balachandran Manavalan","doi":"10.2174/1574893618666230905102407","DOIUrl":"https://doi.org/10.2174/1574893618666230905102407","url":null,"abstract":"\u0000\u0000Krüppel-like factors (KLFs) are a family of transcription factors containing zinc fingers that regulate various cellular processes. KLF proteins are associated with human diseases, such as cancer, cardiovascular diseases, and metabolic disorders. The KLF family consists of 18 members with diverse expression profiles across numerous tissues. Accurate identification and annotation of KLF proteins is crucial, given their involvement in important biological functions. Although experimental approaches can identify KLF proteins precisely, large-scale identification is complicated, slow, and expensive.\u0000\u0000\u0000\u0000In this study, we developed RDR100, a novel random forest (RF)-based framework for predicting KLF proteins based on their primary sequences. First, we identified the optimal encodings for ten different features using a recursive feature elimination approach, and then trained their respective model using five distinct machine learning (ML) classifiers. Results: The performance of all models was assessed using independent datasets, and RDR100 was selected as the final model based on its consistent performance in cross-validation and independent evaluation.\u0000\u0000\u0000\u0000Our results demonstrate that RDR100 is a robust predictor of KLF proteins. RDR100 web server is available at https://procarb.org/RDR100/.\u0000","PeriodicalId":10801,"journal":{"name":"Current Bioinformatics","volume":" ","pages":""},"PeriodicalIF":4.0,"publicationDate":"2023-09-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"48427754","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-09-05DOI: 10.2174/1574893618666230905154441
Ezra B. Wijaya, Venugopala Reddy Mekala, Efendi Zaenudin, Ka-Lok Ng
Background: Metastasis involves multiple stages and various genetic and epigenetic alterations. MicroRNA has been investigated as a biomarker and prognostic tool in various cancer types and stages. Nevertheless, exploring the role of miRNA in kidney cancer remains a significant challenge, given the ability of a single miRNA to target multiple genes within biological networks and pathways. background: Metastasis involves multiple stages and various genetic and epigenetic alterations. MicroRNA has been investigated as a biomarker and prognostic tool in various cancer types and stages. Nevertheless, exploring the role of miRNA in kidney cancer remains a significant challenge, given the ability of a single miRNA to target multiple genes within biological networks and pathways. Objective: This study aims to propose a computational research framework that hypothesizes that a set of miRNAs functions as key regulators in modulating gene expression networks of kidney cancer survival. Method: We retrieved the NGS data from the TCGA-KIRC extracted from UCSC Xena. A set of prognostic miRNAs was acquired through multiple Cox regression analyses. We adopted machine learning approaches to evaluate miRNA prognosis's classification performance between normal, primary (M0), and metastasis (M1) samples. The molecular mechanism between primary cancer and metastasis was investigated by identifying the regulatory networks of miRNA's target genes. Result: A total of 14 miRNAs were identified as potential prognostic indicators. A combination of high-expression miRNAs was associated with survival probability. Machine learning achieved an average accuracy of 95% in distinguishing primary cancer from normal tissue and 79% in predicting the metastasis from primary tissue. Correlation analysis of miRNA prognostics with target genes unveiled regulatory network disparities between metastatic and primary tissues. Conclusion: This study has identified 14 miRNAs that could potentially serve as vital biomarkers for diagnosing and prognosing ccRCC. Differential regulatory networks between metastatic and primary tissues in this study provide the molecular basis for assessment and therapeutic treatment for ccRCC patients
{"title":"In Silico Study of Clinical Prognosis Associated MicroRNAs for patients with Metastasis in Clear Cell Renal Carcinoma","authors":"Ezra B. Wijaya, Venugopala Reddy Mekala, Efendi Zaenudin, Ka-Lok Ng","doi":"10.2174/1574893618666230905154441","DOIUrl":"https://doi.org/10.2174/1574893618666230905154441","url":null,"abstract":"Background: Metastasis involves multiple stages and various genetic and epigenetic alterations. MicroRNA has been investigated as a biomarker and prognostic tool in various cancer types and stages. Nevertheless, exploring the role of miRNA in kidney cancer remains a significant challenge, given the ability of a single miRNA to target multiple genes within biological networks and pathways. background: Metastasis involves multiple stages and various genetic and epigenetic alterations. MicroRNA has been investigated as a biomarker and prognostic tool in various cancer types and stages. Nevertheless, exploring the role of miRNA in kidney cancer remains a significant challenge, given the ability of a single miRNA to target multiple genes within biological networks and pathways. Objective: This study aims to propose a computational research framework that hypothesizes that a set of miRNAs functions as key regulators in modulating gene expression networks of kidney cancer survival. Method: We retrieved the NGS data from the TCGA-KIRC extracted from UCSC Xena. A set of prognostic miRNAs was acquired through multiple Cox regression analyses. We adopted machine learning approaches to evaluate miRNA prognosis's classification performance between normal, primary (M0), and metastasis (M1) samples. The molecular mechanism between primary cancer and metastasis was investigated by identifying the regulatory networks of miRNA's target genes. Result: A total of 14 miRNAs were identified as potential prognostic indicators. A combination of high-expression miRNAs was associated with survival probability. Machine learning achieved an average accuracy of 95% in distinguishing primary cancer from normal tissue and 79% in predicting the metastasis from primary tissue. Correlation analysis of miRNA prognostics with target genes unveiled regulatory network disparities between metastatic and primary tissues. Conclusion: This study has identified 14 miRNAs that could potentially serve as vital biomarkers for diagnosing and prognosing ccRCC. Differential regulatory networks between metastatic and primary tissues in this study provide the molecular basis for assessment and therapeutic treatment for ccRCC patients","PeriodicalId":10801,"journal":{"name":"Current Bioinformatics","volume":"50 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-09-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"135362419","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
As we all know, finding new pharmaceuticals requires a lot of time and money, which has compelled people to think about adopting more effective approaches to locate drugs. Researchers have made significant progress recently when it comes to using Deep Learning (DL) to create DTI.. Therefore, we propose a deep learning model that applies Transformer to DTI prediction. The model uses a Transformer and Graph Transformer to extract the feature information of protein and compound molecules, respectively, and combines their respective representations to predict interactions. We used Human and C.elegans, the two benchmark datasets, evaluated the proposed method in different experimental settings and compared it with the latest DL model. The results show that the proposed model based on DL is an effective method for the classification and recognition of DTI prediction, and its performance on the two data sets is significantly better than other DL based methods.
{"title":"Transformer and graph transformer-based prediction of drug-target interactions","authors":"Weizhong Lu, Meiling Qian, Yu Zhang, Junkai Liu, Hongjie Wu, Yaoyao Lu, Haiou Li, Qiming Fu, Jiyun Shen, Yongbiao Xiao","doi":"10.2174/1574893618666230825121841","DOIUrl":"https://doi.org/10.2174/1574893618666230825121841","url":null,"abstract":"\u0000\u0000As we all know, finding new pharmaceuticals requires a lot of time and money, which has compelled people to think about adopting more effective approaches to locate drugs. Researchers have made significant progress recently when it comes to using Deep Learning (DL) to create DTI..\u0000\u0000\u0000\u0000Therefore, we propose a deep learning model that applies Transformer to DTI prediction. The model uses a Transformer and Graph Transformer to extract the feature information of protein and compound molecules, respectively, and combines their respective representations to predict interactions.\u0000\u0000\u0000\u0000We used Human and C.elegans, the two benchmark datasets, evaluated the proposed method in different experimental settings and compared it with the latest DL model.\u0000\u0000\u0000\u0000The results show that the proposed model based on DL is an effective method for the classification and recognition of DTI prediction, and its performance on the two data sets is significantly better than other DL based methods.\u0000","PeriodicalId":10801,"journal":{"name":"Current Bioinformatics","volume":" ","pages":""},"PeriodicalIF":4.0,"publicationDate":"2023-08-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"43591358","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-08-18DOI: 10.2174/1574893618666230818121046
Rania Hamdy, Yasser M. K. Omar, F. Maghraby
Histone modification is a vital element in gene expression regulation. The way in which these proteins bind to the DNA impacts whether or not a gene may be expressed. Although those factors cannot in-fluence DNA construction, they can influence how it is transcribed. Each spatial location in DNA has its function, so the spatial ar-rangement of chromatin modifications affects how the gene can express. Al-so, gene regulation is affected by the type of histone modification combina-tions that are present on the gene and depends on the spatial distributional pattern of these modifications and how long these modifications read on a gene region. So, this study aims to know how to model Long-range spatial genome data and model complex dependencies among Histone reads. The Convolution Neural Network (CNN) is used to model all da-ta features in this paper. It can detect patterns in histones signals and pre-serve the spatial information of these patterns. It also uses the concept of memory in long short-term memory (LSTM), using vanilla LSTM, Bi-Directional LSTM, or Stacked LSTM to preserve long-range histones sig-nals. Additionally, it tries to combine these methods using ConvLSTM or uses them together with the aid of a self-attention. Based on the results, the combination of CNN, LSTM with the self-attention mechanism obtained an Area under the Curve (AUC) score of 88.87% over 56 cell types. The result outperforms the present state-of-the-art model and provides insight into how combinatorial interactions between histone modi-fication marks can control gene expression. The source code is available at https://github.com/RaniaHamdy/DeepEpi.
组蛋白修饰是基因表达调控的重要组成部分。这些蛋白质与DNA结合的方式影响着基因是否可以表达。尽管这些因素不能影响DNA的构建，但它们可以影响DNA的转录方式。DNA中的每个空间位置都有其功能，因此染色质修饰的空间排列影响基因的表达方式。同样，基因调控受到存在于基因上的组蛋白修饰组合类型的影响，并取决于这些修饰的空间分布模式以及这些修饰在基因区域上读取的时间长短。因此，本研究旨在了解如何对远程空间基因组数据进行建模，并对组蛋白reads之间的复杂依赖关系进行建模。本文使用卷积神经网络(CNN)对所有数据特征进行建模。它可以检测组蛋白信号中的模式，并预先保存这些模式的空间信息。它还使用长短期记忆(LSTM)的概念，使用vanilla LSTM、双向LSTM或堆叠LSTM来保存远程组蛋白信号。此外，它尝试使用ConvLSTM将这些方法结合起来，或者在自我关注的帮助下将它们一起使用。结果表明，CNN、LSTM结合自注意机制在56种细胞类型中获得了88.87%的曲线下面积(Area under the Curve, AUC)得分。结果优于目前最先进的模型，并提供了洞察组蛋白修饰标记之间的组合相互作用如何控制基因表达。源代码可从https://github.com/RaniaHamdy/DeepEpi获得。
{"title":"DeepEpi: Deep Learning Model for Predicting Gene Expression Regulation based on Epigenetic Histone Modifications","authors":"Rania Hamdy, Yasser M. K. Omar, F. Maghraby","doi":"10.2174/1574893618666230818121046","DOIUrl":"https://doi.org/10.2174/1574893618666230818121046","url":null,"abstract":"\u0000\u0000Histone modification is a vital element in gene expression regulation. The way in which these proteins bind to the DNA impacts whether or not a gene may be expressed. Although those factors cannot in-fluence DNA construction, they can influence how it is transcribed.\u0000\u0000\u0000\u0000Each spatial location in DNA has its function, so the spatial ar-rangement of chromatin modifications affects how the gene can express. Al-so, gene regulation is affected by the type of histone modification combina-tions that are present on the gene and depends on the spatial distributional pattern of these modifications and how long these modifications read on a gene region. So, this study aims to know how to model Long-range spatial genome data and model complex dependencies among Histone reads.\u0000\u0000\u0000\u0000The Convolution Neural Network (CNN) is used to model all da-ta features in this paper. It can detect patterns in histones signals and pre-serve the spatial information of these patterns. It also uses the concept of memory in long short-term memory (LSTM), using vanilla LSTM, Bi-Directional LSTM, or Stacked LSTM to preserve long-range histones sig-nals. Additionally, it tries to combine these methods using ConvLSTM or uses them together with the aid of a self-attention.\u0000\u0000\u0000\u0000Based on the results, the combination of CNN, LSTM with the self-attention mechanism obtained an Area under the Curve (AUC) score of 88.87% over 56 cell types.\u0000\u0000\u0000\u0000The result outperforms the present state-of-the-art model and provides insight into how combinatorial interactions between histone modi-fication marks can control gene expression. The source code is available at https://github.com/RaniaHamdy/DeepEpi.\u0000","PeriodicalId":10801,"journal":{"name":"Current Bioinformatics","volume":" ","pages":""},"PeriodicalIF":4.0,"publicationDate":"2023-08-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"43840967","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-08-16DOI: 10.2174/1574893618666230816090443
Jian-ping Zhao, Bo Yang, Hai-yun Wang, Chunhan Zheng
Single-cell RNA sequencing(scRNA-seq) data can identify heterogeneity between cells, thereby identifying cell types and discovering rare cell types. Clustering is often used to identify cell types, but the high noise and high dimension of scRNA-seq lead to the degradation of clustering performance and impact downstream analysis. Deep learning is widely used in this field, which provides promising performance in feature learning. Most deep learning models only consider the relationship between genes, ignore the relationship between cells. We try to use the relationships between cells and the relationships between genes to construct clustering models. We proposed scSDSC: a deep subspace cluster architecture that considers the relationships between genes and cells at the same time. Similar to deep subspace clustering (DSC), we added a fully connected layer after the embedding layer to obtain the self-expression matrix. In addition, we also added a fully connected SoftMax layer to generate the pseudo-label and used the information carried by the pseudo-label for model training. Finally, the affinity matrix is obtained for spectral clustering. Experimental results on eight real datasets show that scSDSC outperforms existing methods in downstream analysis. Our method plays an important role in improving clustering accuracy and downstream analysis.
{"title":"scSDSC: Self-supervised Deep Subspace Clustering for scRNA-seq Data","authors":"Jian-ping Zhao, Bo Yang, Hai-yun Wang, Chunhan Zheng","doi":"10.2174/1574893618666230816090443","DOIUrl":"https://doi.org/10.2174/1574893618666230816090443","url":null,"abstract":"\u0000\u0000Single-cell RNA sequencing(scRNA-seq) data can identify heterogeneity between cells, thereby identifying cell types and discovering rare cell types. Clustering is often used to identify cell types, but the high noise and high dimension of scRNA-seq lead to the degradation of clustering performance and impact downstream analysis. Deep learning is widely used in this field, which provides promising performance in feature learning.\u0000\u0000\u0000\u0000Most deep learning models only consider the relationship between genes, ignore the relationship between cells. We try to use the relationships between cells and the relationships between genes to construct clustering models.\u0000\u0000\u0000\u0000We proposed scSDSC: a deep subspace cluster architecture that considers the relationships between genes and cells at the same time. Similar to deep subspace clustering (DSC), we added a fully connected layer after the embedding layer to obtain the self-expression matrix. In addition, we also added a fully connected SoftMax layer to generate the pseudo-label and used the information carried by the pseudo-label for model training. Finally, the affinity matrix is obtained for spectral clustering.\u0000\u0000\u0000\u0000Experimental results on eight real datasets show that scSDSC outperforms existing methods in downstream analysis.\u0000\u0000\u0000\u0000Our method plays an important role in improving clustering accuracy and downstream analysis.\u0000","PeriodicalId":10801,"journal":{"name":"Current Bioinformatics","volume":" ","pages":""},"PeriodicalIF":4.0,"publicationDate":"2023-08-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"43254950","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-08-16DOI: 10.2174/1574893618666230816141723
Xiaoqing Peng, Wanxin Cui, Wenjin Zhang, Zihao Li, Xiaoshu Zhu, L. Yuan, Ji Li
Identifying differentially methylated region (DMR) is a basic but important task in epigenomics, which can help investigate the mechanisms of diseases and provide methylation biomarkers for screening diseases. A set of methods have been proposed to identify DMRs from methylation array data. However, it lacks effective metrics to characterize different DMR sets and enable a straight way for comparison. In this study, we introduce a metric, DMRn, to characterize DMR sets detected by different methods from methylation array data. To calculate DMRn, firstly, the methylation differences of DMRs are recalculated by incorporating the correlations between probes and their represented CpGs. Then, DMRn is calculated based on the number of probes and the dense of CpGs in DMRs with methylation differences falling in each interval. By comparing the DMRn of DMR sets predicted by seven methods on four scenario, the results demonstrate that DMRn can make an efficient guidance for selecting DMR sets, and provide new insights in cancer genomics studies by comparing the DMR sets from the related pathological states. For example, there are many regions with subtle methylation alteration in subtypes of prostate cancer are altered oppositely in the benign state, which may indicate a possible revision mechanism in benign prostate cancer. Futhermore, when applied to datasets that underwent different runs of batch effect removal, the DMRn can help to visualize the bias introduced by multi-runs of batch effect removal. The tool for calculating DMRn is available in the GitHub repository(https://github.com/xqpeng/DMRArrayMetric).
{"title":"A Metric to Characterize Differentially Methylated Region Sets Detected from Methylation Array Data","authors":"Xiaoqing Peng, Wanxin Cui, Wenjin Zhang, Zihao Li, Xiaoshu Zhu, L. Yuan, Ji Li","doi":"10.2174/1574893618666230816141723","DOIUrl":"https://doi.org/10.2174/1574893618666230816141723","url":null,"abstract":"\u0000\u0000Identifying differentially methylated region (DMR) is a basic but important task in epigenomics, which can help investigate the mechanisms of diseases and provide methylation biomarkers for screening diseases. A set of methods have been proposed to identify DMRs from methylation array data. However, it lacks effective metrics to characterize different DMR sets and enable a straight way for comparison.\u0000\u0000\u0000\u0000In this study, we introduce a metric, DMRn, to characterize DMR sets detected by different methods from methylation array data. To calculate DMRn, firstly, the methylation differences of DMRs are recalculated by incorporating the correlations between probes and their represented CpGs. Then, DMRn is calculated based on the number of probes and the dense of CpGs in DMRs with methylation differences falling in each interval.\u0000\u0000\u0000\u0000By comparing the DMRn of DMR sets predicted by seven methods on four scenario, the results demonstrate that DMRn can make an efficient guidance for selecting DMR sets, and provide new insights in cancer genomics studies by comparing the DMR sets from the related pathological states. For example, there are many regions with subtle methylation alteration in subtypes of prostate cancer are altered oppositely in the benign state, which may indicate a possible revision mechanism in benign prostate cancer.\u0000\u0000\u0000\u0000Futhermore, when applied to datasets that underwent different runs of batch effect removal, the DMRn can help to visualize the bias introduced by multi-runs of batch effect removal. The tool for calculating DMRn is available in the GitHub repository(https://github.com/xqpeng/DMRArrayMetric).\u0000","PeriodicalId":10801,"journal":{"name":"Current Bioinformatics","volume":" ","pages":""},"PeriodicalIF":4.0,"publicationDate":"2023-08-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"42313996","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-08-16DOI: 10.2174/1574893618666230816090548
Yaoyao Lu, Junkai Liu, T. Jiang, Zhiming Cui, Hongjie Wu
New drugs are costly, time-consuming, and often accompanied by safety concerns. With the development of deep learning, computer-aided drug design has become more mainstream, and convolutional neural networks and graph neural networks have been widely used for drug–target affinity (DTA) prediction. The paper proposes a method of predicting DTA using graph convolutional networks and multiscale convolutional neural networks. We construct drug molecules into graph representation vectors and learn feature expressions through graph attention networks and graph convolutional networks. A three-branch convolutional neural network learns the local and global features of protein sequences, and the two feature representations are merged into a regression module to predict the DTA. We present a novel model to predict DTA, with a 2.5% improvement in the consistency index and a 21% accuracy improvement in terms of the mean squared error on the Davis dataset compared to DeepDTA. Morever, our method outperformed other mainstream DTA prediction models namely, GANsDTA, WideDTA, GraphDTA and DeepAffinity. The results showed that the use of multiscale convolutional neural networks was better than a single-branched convolutional neural network at capturing protein signatures and the use of graphs to express drug molecules yielded better results.
{"title":"Drug–target binding affinity prediction based on three-branched multiscale convolutional neural networks","authors":"Yaoyao Lu, Junkai Liu, T. Jiang, Zhiming Cui, Hongjie Wu","doi":"10.2174/1574893618666230816090548","DOIUrl":"https://doi.org/10.2174/1574893618666230816090548","url":null,"abstract":"\u0000\u0000New drugs are costly, time-consuming, and often accompanied by safety concerns. With the development of deep learning, computer-aided drug design has become more mainstream, and convolutional neural networks and graph neural networks have been widely used for drug–target affinity (DTA) prediction.\u0000\u0000\u0000\u0000The paper proposes a method of predicting DTA using graph convolutional networks and multiscale convolutional neural networks.\u0000\u0000\u0000\u0000We construct drug molecules into graph representation vectors and learn feature expressions through graph attention networks and graph convolutional networks. A three-branch convolutional neural network learns the local and global features of protein sequences, and the two feature representations are merged into a regression module to predict the DTA.\u0000\u0000\u0000\u0000We present a novel model to predict DTA, with a 2.5% improvement in the consistency index and a 21% accuracy improvement in terms of the mean squared error on the Davis dataset compared to DeepDTA. Morever, our method outperformed other mainstream DTA prediction models namely, GANsDTA, WideDTA, GraphDTA and DeepAffinity.\u0000\u0000\u0000\u0000The results showed that the use of multiscale convolutional neural networks was better than a single-branched convolutional neural network at capturing protein signatures and the use of graphs to express drug molecules yielded better results.\u0000","PeriodicalId":10801,"journal":{"name":"Current Bioinformatics","volume":" ","pages":""},"PeriodicalIF":4.0,"publicationDate":"2023-08-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"44958222","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-08-15DOI: 10.2174/1574893618666230815121150
Leying Pan, T. Zhang, Qiang Yang, Guoping Yang, Nan Han, Shaojie Qiao
Breast tumor is among the most malignant tumors and early detection can improve patient’s survival rate. Currently, mammography is the most reliable method for diagnosing breast tumor because of high image resolution. Because of the rapid development of medical and artificial intelligence techniques, computer-aided diagnosis technology can greatly improve the detection accuracy of breast tumors and medical imaging has begun to use deep-learning-based approaches. In this study, the TumorDet model is proposed to detect the benign and malignant lesions of breast tumor, which has positive significance for assisting doctors in diagnosis. We use the proposed TumorDet to analyze and predict breast tumors on the real MRI dataset. (1) We introduce an adaptive gamma correction (AGC) method to balance brightness equalization and increase the contrast of mammography images; (2) we use the ShuffleNet model to exchange information between different feature layers and extract the hidden high-level features of medical images; and (3) we use the transfer learning method to fine-tune the ShuffleNet model and obtain the optimal parameters. The proposed TumorDet model has shown that accuracy, sensitivity, and specificity reach 90.43%, 89.37%, and 87.81%, respectively. TumorDet performs well in the breast tumor detection task. In addition, we use the proposed TumorDet to conduct experiments on other tasks, such as forest fires, and the robustness of TumorDet is proved by experimental results. TumorDet employs the ShuffleNet model to exchange information between different feature layers without increasing the number of network parameters and applies transfer learning methods to further extract the basic features of medical images by fine-tuning. The model is beneficial for the localization and classification of breast tumors and also performs well in forest fire detection.
{"title":"TumorDet: A Breast Tumor Detection Model Based on Transfer Learning and ShuffleNet","authors":"Leying Pan, T. Zhang, Qiang Yang, Guoping Yang, Nan Han, Shaojie Qiao","doi":"10.2174/1574893618666230815121150","DOIUrl":"https://doi.org/10.2174/1574893618666230815121150","url":null,"abstract":"\u0000\u0000Breast tumor is among the most malignant tumors and early detection can improve patient’s survival rate. Currently, mammography is the most reliable method for diagnosing breast tumor because of high image resolution. Because of the rapid development of medical and artificial intelligence techniques, computer-aided diagnosis technology can greatly improve the detection accuracy of breast tumors and medical imaging has begun to use deep-learning-based approaches. In this study, the TumorDet model is proposed to detect the benign and malignant lesions of breast tumor, which has positive significance for assisting doctors in diagnosis.\u0000\u0000\u0000\u0000We use the proposed TumorDet to analyze and predict breast tumors on the real MRI dataset.\u0000\u0000\u0000\u0000(1) We introduce an adaptive gamma correction (AGC) method to balance brightness equalization and increase the contrast of mammography images; (2) we use the ShuffleNet model to exchange information between different feature layers and extract the hidden high-level features of medical images; and (3) we use the transfer learning method to fine-tune the ShuffleNet model and obtain the optimal parameters.\u0000\u0000\u0000\u0000The proposed TumorDet model has shown that accuracy, sensitivity, and specificity reach 90.43%, 89.37%, and 87.81%, respectively. TumorDet performs well in the breast tumor detection task. In addition, we use the proposed TumorDet to conduct experiments on other tasks, such as forest fires, and the robustness of TumorDet is proved by experimental results.\u0000\u0000\u0000\u0000TumorDet employs the ShuffleNet model to exchange information between different feature layers without increasing the number of network parameters and applies transfer learning methods to further extract the basic features of medical images by fine-tuning. The model is beneficial for the localization and classification of breast tumors and also performs well in forest fire detection.\u0000","PeriodicalId":10801,"journal":{"name":"Current Bioinformatics","volume":" ","pages":""},"PeriodicalIF":4.0,"publicationDate":"2023-08-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"49632050","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-08-07DOI: 10.2174/1574893618666230807123016
Baoping Zhu, Huizhen Geng, Fan Yang, Yanxin Wu, Tiefeng Cao, Dongyu Wang, Zilian Wang
Preeclampsia (PE) is a severe pregnancy complication associated with autophagy. This research sought to uncover autophagy-related genes in pre-eclampsia through bioinformatics and machine learning. GSE75010 from the GEO series was subjected to WGCNA to identify key modular genes in PE. Autophagy genes retrieved from the THANATOS overlapped with the modular genes to yield PE-related autophagy genes. Furthermore, the crucial step involved the utilization of two machine learning algorithms (LASSO and SVM-RFE) for dimensionality reduction. The candidate gene was further verified by quantitative reverse transcription polymerase chain reaction, western blot, and immunohistochemistry. Preliminary experiments were conducted on HTR-8/SVneo cell lines to explore the role of candidate genes in autophagy regulation. WGCNA identified 291 genes from 5 hubs, and after overlapping with 1087 autophagy-related genes obtained from THANATOS, 42 PE-related ARGs were identified. ANXA6 was recognized as a potential target through SVM-RFE and LASSO analyses. The mRNA and protein expression of ANXA6 were verified in placenta samples. In HTR8/SVneo cells, modulating ANXA6 expression altered autophagy levels. Knocking down ANXA6 resulted in an anti-autophagy effect, which was reversed by treatment with CAL101, an inhibitor of PI3K, Akt, and mTOR. We observed that ANXA6 may serve as a possible PE action target and that autophagy may be crucial to the pathogenesis of PE.
{"title":"Revealing ANXA6 as a Novel Autophagy-related Target for Pre-eclampsia Based on the Machine Learning","authors":"Baoping Zhu, Huizhen Geng, Fan Yang, Yanxin Wu, Tiefeng Cao, Dongyu Wang, Zilian Wang","doi":"10.2174/1574893618666230807123016","DOIUrl":"https://doi.org/10.2174/1574893618666230807123016","url":null,"abstract":"\u0000\u0000Preeclampsia (PE) is a severe pregnancy complication associated with autophagy.\u0000\u0000\u0000\u0000This research sought to uncover autophagy-related genes in pre-eclampsia through bioinformatics and machine learning.\u0000\u0000\u0000\u0000GSE75010 from the GEO series was subjected to WGCNA to identify key modular genes in PE. Autophagy genes retrieved from the THANATOS overlapped with the modular genes to yield PE-related autophagy genes. Furthermore, the crucial step involved the utilization of two machine learning algorithms (LASSO and SVM-RFE) for dimensionality reduction. The candidate gene was further verified by quantitative reverse transcription polymerase chain reaction, western blot, and immunohistochemistry. Preliminary experiments were conducted on HTR-8/SVneo cell lines to explore the role of candidate genes in autophagy regulation.\u0000\u0000\u0000\u0000WGCNA identified 291 genes from 5 hubs, and after overlapping with 1087 autophagy-related genes obtained from THANATOS, 42 PE-related ARGs were identified. ANXA6 was recognized as a potential target through SVM-RFE and LASSO analyses. The mRNA and protein expression of ANXA6 were verified in placenta samples. In HTR8/SVneo cells, modulating ANXA6 expression altered autophagy levels. Knocking down ANXA6 resulted in an anti-autophagy effect, which was reversed by treatment with CAL101, an inhibitor of PI3K, Akt, and mTOR.\u0000\u0000\u0000\u0000We observed that ANXA6 may serve as a possible PE action target and that autophagy may be crucial to the pathogenesis of PE.\u0000","PeriodicalId":10801,"journal":{"name":"Current Bioinformatics","volume":" ","pages":""},"PeriodicalIF":4.0,"publicationDate":"2023-08-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"45513523","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
RNA editing enriches post-transcriptional sequence changes. Currently detecting RNA editing sites is mostly based on the Sanger sequencing platform and second-generation sequencing. However, detection with Sanger sequencing is limited by the disturbing background peaks using the direct sequencing method and the clone number using the clone sequencing method, while second-generation sequencing detection is constrained by its short read. We aimed to design a pipeline that can accurately detect RNA editing sites for full-length long-read amplicons to meet the requirement when focusing on a few specific genes of interest. We developed a novel high-throughput RNA editing sites detection pipeline based on the PacBio circular consensus sequences sequencing which is accurate with high-throughput and long-read coverage. We tested the pipeline on cytosolic malate dehydrogenase in the hard-shelled mussel Mytilus coruscus and further validated it using direct Sanger sequencing. Data generated from the PacBio circular consensus sequences (CCS) amplicons in three mussels were first filtered by quality and then selected by open reading frame. After filtering, 225-2047 sequences of the three mussels, respectively, were used to identify RNA editing sites. With corresponding genomic DNA sequences, we extracted 227-799 candidate RNA editing sites excluding heterozygous sites. We further figured out 7-11 final RESs using a new error model specially designed for RNA editing site detection. The resulting RNA editing sites all agree with the validation using the Sanger sequencing. We report a near-zero error rate method in identifying RNA editing sites of long-read amplicons with the use of PacBio CCS sequencing.
{"title":"Full-length PacBio Amplicon Sequencing to Unveil RNA Editing Sites","authors":"Xiao-lu Zhu, Ming-ling Liao, Ya-Jie Zhu, Yun‐wei Dong","doi":"10.2174/1574893618666230803112142","DOIUrl":"https://doi.org/10.2174/1574893618666230803112142","url":null,"abstract":"\u0000\u0000RNA editing enriches post-transcriptional sequence changes. Currently detecting RNA editing sites is mostly based on the Sanger sequencing platform and second-generation sequencing. However, detection with Sanger sequencing is limited by the disturbing background peaks using the direct sequencing method and the clone number using the clone sequencing method, while second-generation sequencing detection is constrained by its short read.\u0000\u0000\u0000\u0000We aimed to design a pipeline that can accurately detect RNA editing sites for full-length long-read amplicons to meet the requirement when focusing on a few specific genes of interest.\u0000\u0000\u0000\u0000We developed a novel high-throughput RNA editing sites detection pipeline based on the PacBio circular consensus sequences sequencing which is accurate with high-throughput and long-read coverage. We tested the pipeline on cytosolic malate dehydrogenase in the hard-shelled mussel Mytilus coruscus and further validated it using direct Sanger sequencing.\u0000\u0000\u0000\u0000Data generated from the PacBio circular consensus sequences (CCS) amplicons in three mussels were first filtered by quality and then selected by open reading frame. After filtering, 225-2047 sequences of the three mussels, respectively, were used to identify RNA editing sites. With corresponding genomic DNA sequences, we extracted 227-799 candidate RNA editing sites excluding heterozygous sites. We further figured out 7-11 final RESs using a new error model specially designed for RNA editing site detection. The resulting RNA editing sites all agree with the validation using the Sanger sequencing.\u0000\u0000\u0000\u0000We report a near-zero error rate method in identifying RNA editing sites of long-read amplicons with the use of PacBio CCS sequencing.\u0000","PeriodicalId":10801,"journal":{"name":"Current Bioinformatics","volume":" ","pages":""},"PeriodicalIF":4.0,"publicationDate":"2023-08-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"45187580","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：