請用此 Handle URI 來引用此文件:
http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/74090
標題: | 應用深度神經網路進行基於單一核苷酸多型性的DNA混合物解譯 Applying Deep Neural Networks for SNP-based DNA Mixture Interpretation |
作者: | Yi-Hao Li 李奕皓 |
指導教授: | 周承復(Cheng-Fu Chou) |
關鍵字: | 機器學習,深度學習,神經網路,單一核?酸多型性,DNA混合物, machine learning,deep learning,neural networks,single nucleotide polymorphism (SNP),DNA mixture, |
出版年 : | 2019 |
學位: | 碩士 |
摘要: | 單一核苷酸多型性 (single nucleotide polymorphism; SNP) 被廣泛用於臨床研究和法醫學應用,然而對於DNA混合物解譯 (DNA mixture interpretation),基於短縱列重複序列 (short tandem repeat; STR) 的方法占主導地位。本論文的目的是探索深度神經網路 (deep neural networks) 可以幫助基於SNP的DNA混合物解譯的可能性。為此,我們提出了兩種解決方案:一種是基於線性迴歸,另一種是基於深度神經網路。如結果所示,深度神經網路在解譯DNA混合物上具有優勢。儘管裂解的檢體以及具血緣關係個體的DNA混合物在解譯上仍然具有挑戰性,但神經網路具有更高的DNA混合物解譯準確度。基於該結果,我們得出結論:有了深度神經網路,DNA混合物解譯不必被限制於基於STR的方法。 Single nucleotide polymorphisms (SNPs) have been widely used in clinical research and forensic applications while, for the DNA mixture interpretation, the methods based on short tandem repeats (STRs) are dominating. The goal of this thesis is to explore the possibility that deep neural networks can help with the SNP-based DNA mixture interpretation. For this purpose, we proposed two solutions: one is based on linear regression and the other is based on deep neural networks. As demonstrated in the result, the deep neural networks exhibit the advantages of interpreting DNA mixtures. Though the interpretation of degraded samples and the mixtures contributed by blood-related individuals are still challenging, the neural networks have higher accuracy of the DNA mixture interpretation. Based on the result, we conclude that, with deep neural networks, the DNA mixture interpretation is not limited to STR-based methods. |
URI: | http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/74090 |
DOI: | 10.6342/NTU201903533 |
全文授權: | 有償授權 |
顯示於系所單位: | 資訊工程學系 |
文件中的檔案:
檔案 | 大小 | 格式 | |
---|---|---|---|
ntu-108-1.pdf 目前未授權公開取用 | 1.07 MB | Adobe PDF |
系統中的文件,除了特別指名其著作權條款之外,均受到著作權保護,並且保留所有的權利。