lncRNA基因调控关系的分析与预测
发布时间:2021-01-12 10:19
研究表明,每种非编码RNA(ncRNA)不仅可以通过其靶基因起作用,而且可以彼此相互作用以作用影响于生物学性状,并且这种相互作用更常见。许多研究主要集中在微小RNA(miRNA)和信使RNA(mRNA)相互作用的分析上。本研究中,提出了两个独立的模型来分析和预测lncRNA基因调控关系。第一个模型基于传统支持向量回归(SVR),第二类模型基于深度集成学习。在第一个模型中,使用SVR研究了拟南芥miRNA和长非编码RNA(lncRNA)相互作用,模型可以识别出新的相互作用并分析在胁迫响应下的调节作用。构建并分析了miRNAmRNA,miRNA-lncRNA和miRNA-mRNA-lncRNA的互作网络。我们发现具有低序列号的miRNA,具有高序列号的靶向lncRNA和具有高序列号的miRNA靶向具有低序列号的lncRNA。实验结果表明miRNA-lncRNA之间存在调节关系。使用具有新基因表达机制的SVR预测新RNA靶标,并标注了胁迫响应相关功能。在第二个模型中,我们使用长短期记忆自动编码器(LSTM-AE)在相同的数据集上研究了miRNA-lncRNA序列的相互作用。实验结果表明,方法...
【文章来源】:大连理工大学辽宁省 211工程院校 985工程院校 教育部直属院校
【文章页数】:48 页
【学位级别】:硕士
【文章目录】:
Abstract
摘要
1 Introduction
1.1 Related Work and significance
1.2 Domestic and Overseas Progress
1.3 Research Content and methodology
1.4 Objectives
1.5 Key Problems solved
2 Methods for Predicting lncRNA-gene Regulatory Relationship
2.1 SVR Based on Traditional SVM
2.1.1 Target Prediction with psRNATarget and TAPIR
2.1.2 RNAs Network Construction
2.1.3 SVR
2.2 LSTM-AE Based on Ensemble Deep Learning
2.2.1 RNA Feature Encoding
2.2.2 Dimensionality Reduction with Auto-Encoders
2.2.3 Data Partitioning:Training,Validation,and Test Sets
2.2.4 LSTM
2.2.5 Stacked LSTM
3 Results
3.1 SVR Based on Traditional SVM
3.1.1 Network Analysis
3.1.2 SVR Approach to Predict miRNA targeting lncRNA
3.1.3 Identifying Regulatory Rules with Stress Response
3.2 LSTM-AE Based on Ensemble Deep Learning
3.2.1 Generating Negative Samples from the Positive Samples
3.2.2 Evaluation of Performance
3.2.3 LSTM-AE
3.3 Comparison of Deep Learning LSTM-AE with Traditional SVR
Conclusion
References
Research Projects and Publications in Master Study
Acknowledgement
本文编号:2972679
【文章来源】:大连理工大学辽宁省 211工程院校 985工程院校 教育部直属院校
【文章页数】:48 页
【学位级别】:硕士
【文章目录】:
Abstract
摘要
1 Introduction
1.1 Related Work and significance
1.2 Domestic and Overseas Progress
1.3 Research Content and methodology
1.4 Objectives
1.5 Key Problems solved
2 Methods for Predicting lncRNA-gene Regulatory Relationship
2.1 SVR Based on Traditional SVM
2.1.1 Target Prediction with psRNATarget and TAPIR
2.1.2 RNAs Network Construction
2.1.3 SVR
2.2 LSTM-AE Based on Ensemble Deep Learning
2.2.1 RNA Feature Encoding
2.2.2 Dimensionality Reduction with Auto-Encoders
2.2.3 Data Partitioning:Training,Validation,and Test Sets
2.2.4 LSTM
2.2.5 Stacked LSTM
3 Results
3.1 SVR Based on Traditional SVM
3.1.1 Network Analysis
3.1.2 SVR Approach to Predict miRNA targeting lncRNA
3.1.3 Identifying Regulatory Rules with Stress Response
3.2 LSTM-AE Based on Ensemble Deep Learning
3.2.1 Generating Negative Samples from the Positive Samples
3.2.2 Evaluation of Performance
3.2.3 LSTM-AE
3.3 Comparison of Deep Learning LSTM-AE with Traditional SVR
Conclusion
References
Research Projects and Publications in Master Study
Acknowledgement
本文编号:2972679
本文链接:https://www.wllwen.com/kejilunwen/zidonghuakongzhilunwen/2972679.html