臺灣博碩士論文加值系統

微陣列為目前癌症研究領域中最常被使用的一個工具，利用其具有容納龐大資料的特性，來記錄基因在癌症之表現，比較正常細胞與癌症細胞之差異。然而在目前的癌症研究上，對於如何分析微陣列資料仍舊未有一個明確的定論。本論文即先以卵巢癌之微陣列資料作為樣本，加以統計方法及演算法建立一個疾病分析模型。首先以主成分分析法對資料進行前置資料的篩選，然後再以變異數分析找出具有差異性的表現基因作為標靶基因，最後以不同的監督式學習方法評估這些基因的分類正確率；若這些基因之正確率高過門檻值，則利用Dijkstra演算法和貝氏定理找出以這些標靶基因建立的調控路徑，以供研究者能夠更了解這些基因的功能與方向性。此分析模型可減少實驗的錯誤嘗試次數及實驗時間，不僅能分析複雜的癌症表現資料，亦可對其他疾病之研究能夠有所幫助。希望未來能夠有效地應用在藥物開發實驗上，來加以找到更有效的治療方法。

In the current cancer research field, microarray is one of the most commonly used tools. It has the advantage of containing a large amount of data, which helped us in recording gene expressions in cancer and comparing the difference between normal cells and cancer cells. However, contemporary cancer research does not have a positive definition in how to analyze the microarray data. In this essay, we utilize the microarray data from the carcinoma cancer as primary sample. And we apply statistical methods and mathematical calculations to establish a diseases analyzing model. At first, we use principal component analysis to process the pre-selected data. Then we use ANOVA to select the genes with significant expression differences to be our target genes. Finally we use supervised learning method to evaluate the accuracy of classification. If the accuracy of target genes is higher than the threshold we set, we apply Dijkstra algorithm and Bayesian theorem to construct the gene regulatory network for these genes. It can provide the researchers with a better understanding in the functionality and regulated directions of these genes. This analytical model helps to reduce the number of incorrect attempts and cut down the time which has to be spent in experiments. This model can analyze complicated cancer expression data, and it can also be useful in researches for other diseases. In prospect, this analytical method can be used in medicine development, for discovering more efficient treatments.

目 錄
論文中文摘要………………………………………………………………………Ⅰ
論文英文摘要…………………………………………............................................Ⅱ
目錄……………………………………………………………………………..….Ⅲ
表目錄………………………………………………………………….…………..Ⅴ
圖目錄………………………………………………………………………………Ⅵ
第一章 緒論…………………………………………………………………………….1
1.1 研究背景與動機…………………………………………………………………1
1.2 研究目的…………………………….………….………………………………..3
1.3 研究方法…………………………….………….………………………………..4
1.3.1 微陣列…………………………….……….………………………………..4
1.3.2 數學模型…………………………….………….…………………………..4
1.4 論文架構…………………………….………….………………………………..5
第二章 文獻探討……………………………………………………………………….6
2.1 人類基因體計畫…………………………………………………………………6
2.2 微陣列分析…….……….………………………………………………………..7
2.3 標靶治療…………………………….………….………………………………..9
2.4 機器學習…………………………….………….………………………………10
2.5 基因調控網路…….…………………………………………………………….11
第三章 研究架構與方法……………………………………………………………...13
3.1 資料處理…………………………….………….………………………………13
3.1.1 主成分分析……………………………………………………..………...15
3.1.2 變異數分析……………………………………………………..………...18
3.2 樣本區分…..…….……………………………………………………………...19
3.2.1 類神經網路……………………………………………………..………...19
3.2.2 貝氏決策樹……………………………………………………..………...23
3.2.3 隨機森林…..……………………………………………………………...26
3.2.4 羅吉斯迴歸……………………………………………………..………...28
3.3調控網路之建立…..…….………………………………………………….........29
3.3.1 Dijkstra演算法…...………………………………………………………..30
3.3.2 貝氏定理..………………………………………………..…………….....32
第四章 研究結果……………………………………………………………………...34
4.1 標靶基因篩選…….…………………………………………………………….34
4.1.1 PCA分析結果..…………………………………………..……………......34
4.1.2 ANOVA分析結果..…………………………...…………..…………….....35
4.2 標靶基因分類結果…….……………………………………………………….38
4.2.1 個別分類結果..………………………………………..…………….........38
4.2.2 實驗總成果..……………..…………………..…………...........................40
4.2.3 系統評估..………………………………………………..…………….....42
4.3 基因調控網路之建立結果…….……………………………………………….44
4.4 基因調控網路之驗證………...…………………………………………..…….51
4.4.1 階層式分群法建立之GPN比較..…………………………………..……52
4.4.2 不同樣本之微陣列資料驗證..……………………….……………..……55
第五章 未來展望……………………………………………………………………...59
參考文獻…………………..…………………………………………………………...61


表 目 錄
表一 基因調控網路建立方法之比較………………………………………………...12
表二 國際婦產科聯盟之卵巢癌分期規定…………………………………………...14
表三 貝氏決策樹分類方法…………………………………………………………...19
表四 Dijkstra演算法執行步驟………………………………………………………32
表五 ANOVA數值分析表(ADCY6)………………………………………………...35
表六 ANOVA數值分析表(PSMB8)………………………………………………...35
表七 以ANOVA篩選的46個標靶基因……………………………………………...41
表八 標靶基因分類結果評估………………………………………………………...43
表九 調控路徑與GeneCards資料庫比對結果………………………………….....46
表十 以階層式分群法所找出的相同路徑…………………………………………...55
表十一 以15個樣本篩選出的16個標靶基因………………………………………55
表十二 15個樣本的標靶基因分類…………………………………………………...56
表十三 調控路徑結果比較…………………………………………………………...59


圖 目 錄
圖1.1 行政院衛生署97年與96年主要死因與死亡人數……………………………..1
圖1.2 細胞週期示意圖…………………………………………………………………2
圖2.1 中心法則…………………………………………………………………………6
圖2.2 cDNA晶片與寡核甘酸晶片製程圖……...……………………………………8
圖2.3 基因調控網路之結構………………………...………………………………...11
圖3.1 研究流程………………………………………………………………………..13
圖3.2 基因表現趨勢…………………………………………………………………..15
圖3.3 PCA降維示意圖……………………………………………………………...16
圖3.4 人類神經元結構………………………..………………………………………19
圖3.5 類神經網路主要架構…………………………………………………………..20
圖3.6 C4.5決策樹建構示意圖…………………………………………….………..24
圖3.7 隨機森林分類步驟……………………………………………………………..27
圖3.8 羅吉斯迴歸分布圖……………………………………………………………..29
圖3.9 Dijkstra演算法範例…………………………………………………………..31
圖4.1 主成分之資料分布圖…………………………………………………………..34
圖4.2 ADCY6(左)與PSMB8(右)在不同樣本變異之盒型圖………………………36
圖4.3 ADCY6多重比較之結果……………………………………………………..37
圖4.4 PSMB8多重比較之結果……………………………………………………...37
圖4.5 不同分類器於不同實驗組的分類結果………………………………………..39
圖4.6 12組標靶基因實驗分類結果………………………………………………...40
圖4.7 基因調控功能分布圖…………………………………………………………..41
圖4.8 不同分類器之AUC分布圖……………………………………………………44
圖4.9 調控網路之距離門檻值分析(41個樣本)……………………………………...44
圖4.10 基因於不同樣本之表現趨勢…..……………………………………………..45
圖4.11 調控網路之資料亂度分析(41個樣本)……………………………..………..46
圖4.12 調控網路建構圖(一)………………………………………………………….48
圖4.13 調控網路建構圖(二)………………………………………………………….49
圖4.14 調控網路建構圖(三)………………………………………………………….50
圖4.15 調控網路建構圖(四)………………………………………………………….51
圖4.16 階層式分群法基本樹狀結構圖………………………………………………52
圖4.17 階層式分群結果………………………………………………………………53
圖4.18 以階層式分群建立之調控網路圖……………………………………………54
圖4.19 調控網路之距離門檻值分析(15個樣本)…………………………………….56
圖4.20 調控網路之資料亂度分析(15個樣本)…..…………………………………..57
圖4.21 以15個樣本建構之調控網路建構圖………………………………………...58

Reference
中文文獻
[1] 行政院衛生署: 96年與97年主要死亡原因與死亡人數.
[2] 台灣癌症基金會副董事長 彭汪嘉康院士,”癌症治療的新思維”,科學人雜誌2003年11月號.
[3] 顏厥全, 癌症與血管新生
[4] 蔡政安, 微陣列資料分析(Microarray analysis)

英文文獻
[5] Jemal, A., Thomas, A., Murray, T., & Thun, M. (2002). Cancer statistics, 2002. Ca-a Cancer Journal for Clinicians, 52(1), 23-47.
[6] Landis, S. H., Murray, T., Bolden, S., & Wingo, P. A. (1999). Cancer statistics, 1999. Ca-a Cancer Journal for Clinicians, 49(1), 8-31.
[7] Ross, D. T., Scherf, U., Eisen, M. B., Perou, C. M., Rees, C., Spellman, P., et al. (2000). Systematic variation in gene expression patterns in human cancer cell lines. Nature Genetics, 24(3), 227-235.
[8] Adams, M. D., Kelley, J. M., Gocayne, J. D., Dubnick, M., Polymeropoulos, M. H., Xiao, H., et al. (1991). Complementary-DNA Sequencing - Expressed Sequence Tags and Human Genome Project. Science, 252(5013), 1651-1656.
[9] Schena, M., Shalon, D., Davis, R. W., & Brown, P. O. (1995). Quantitative Monitoring of Gene-Expression Patterns with a Complementary-DNA Microarray. Science, 270(5235), 467-470.
[10] Schulze, A., & Downward, J. (2001). Navigating gene expression using microarrays - a technology review. Nature Cell Biology, 3(8), E190-E195.
[11] Chou, C. C., Chen, C. H., Lee, T. T., & Peck, K. (2004). Optimization of probe length and the number of probes per gene for optimal microarray analysis of gene expression. Nucleic Acids Research, 32(12), 1-8.
[12] Ross, J. S., Fletcher, J. A., Linette, G. P., Stec, J., Clark, E., Ayers, M., et al. (2003). The HER-2/neu gene and protein in breast cancer 2003: Biomarker and target of therapy. Oncologist, 8(4), 307-325.
[13] Russell, I., & Haller, S. (2003). Tools and techniques of artificial intelligence. International Journal of Pattern Recognition and Artificial Intelligence, 17(5), 685-687.
[14] Ruan, X. G., Wang, J. L., Li, H., Perozzi, R. E., & Perozzi, E. F. (2008). The use of logic relationships to model colon cancer gene expression networks with mRNA microarray data. Journal of Biomedical Informatics, 41(4), 530-543.
[15] Cooke, E. J., Savage, R. S., & Wild, D. L. (2009). Computational approaches to the integration of gene expression, ChIP-chip and sequence data in the inference of gene regulatory networks. Seminars in Cell & Developmental Biology, 20(7), 863-868.
[16] Friedman, N. (2004). Inferring cellular networks using probabilistic graphical models. Science, 303(5659), 799-805.
[17] Shmulevich, I., Dougherty, E. R., Kim, S., & Zhang, W. (2002). Probabilistic Boolean networks: a rule-based uncertainty model for gene regulatory networks. Bioinformatics, 18(2), 261-274.
[18] Yu-Chun Lin, Hsiang-Yuan Yeh, Shih-Wu Cheng, Von-Wun Soo1. (2007). Comparing Cancer and Normal Gene Regulatory Networks Based on Microarray Data and Transcription Factor Analysis. IEEE 2007, 151-157.
[19] Sun, L. F., Jiang, L., Li, M. H., & He, D. C. (2006). Statistical analysis of gene regulatory networks reconstructed from gene expression data of lung cancer. Physica a-Statistical Mechanics and Its Applications, 370(2), 663-671..
[20] Zhou, X. H., Kao, M. C. J., & Wong, W. H. (2002). Transitive functional annotation by shortest-path analysis of gene expression data. Proceedings of the National Academy of Sciences of the United States of America, 99(20), 12783-12788.
[21] Jolliffe I.T. (2002). Principal Component Analysis. Springer Series in Statistics (2nd ed.), 28.
[22] Yeung, K. Y., & Ruzzo, W. L. (2001). Principal component analysis for clustering gene expression data. Bioinformatics, 17(9), 763-774.
[23] Fan, X. T. (1991). Designing Experiments and Analyzing Data - a Model Comparison Perspective - Maxwell,Se, Delaney,Hd. Educational and Psychological Measurement, 51(4), 1063-1069.
[24] Burke H.B., Rosen D.B., Goodman P.H. (1994). Comparing artificial neural networks to other statistical methods for medical outcome prediction. Neural Networks, IEEE International Conference 4, 2213-2216.
[25] Zhu, Y., & Yan, H. (1997). Computerized tumor boundary detection using a Hopfield neural network. Ieee Transactions on Medical Imaging, 16(1), 55-67.
[26] Vlahou, A., Schorge, J. O., Gregory, B. W., & Coleman, R. L. (2003). Diagnosis of ovarian cancer using decision tree classification of mass spectral data. Journal of Biomedicine and Biotechnology(5), 308-314.
[27] Esposito, F., Malerba, D., & Semeraro, G. (1997). A comparative analysis of methods for pruning decision trees. Ieee Transactions on Pattern Analysis and Machine Intelligence, 19(5), 476-491.
[28] Ron Kohavi. (1996). Scaling Up the Accuracy of Naive-Bayes Classifiers: a Decision Tree Hybrid. Proceedings of the Second International Conference on Knowledge Discovery and Data Mining, 202-207.
[29] Leo Breiman (2001) “Random Forest” Machine Learning, 45, 5-32.
[30] Thomas H. Cormen, Charles E. Leiserson, Ronald L. Rivest, and Clifford Stein. Introduction to Algorithms, Second Edition MIT Press and McGraw-Hill, 2001. ISBN 0262032937. Section 24.3: Dijkstra''s algorithm, pp.595-601.
[31] Diamond, G. A., & Forrester, J. S. (1979). Analysis of Probability as an Aid in the Clinical-Diagnosis of Coronary-Artery Disease. New England Journal of Medicine, 300(24), 1350-1358.
[32] Liao, J. G., & Chin, K. V. (2007). Logistic regression for disease classification using microarray data: model selection in a large p and small n case. Bioinformatics, 23(15), 1945-1951.
[33] Werneck, G. L., de Carvalho, D. M., Barroso, D. E., Cook, E. F., & Walker, A. M. (1999). Classification trees and logistic regression applied to prognostic studies: A comparison using meningococcal disease as an example. Journal of Tropical Pediatrics, 45(4), 248-251.
[34] Devoogdt, N., Rasool, N., Hoskins, E., Simpkins, F., Tchabo, N., & Kohn, E. C. (2009). Overexpression of protease inhibitor-dead secretory leukocyte protease inhibitor causes more aggressive ovarian cancer in vitro and in vivo. Cancer Science, 100(3), 434-440.
[35] Hasegawa, S., Miyoshi, Y., Egawa, C., Ishitobi, M., Tamaki, Y., Monden, M., et al. (2002). Mutational analysis of the class I beta-tubulin gene in human breast cancer. International Journal of Cancer, 101(1), 46-51.
[36] Jiang, E. L., He, X., Chen, X. C., Sun, G. J., Wu, H. B., Wei, Y. Q., et al. (2008). Expression of CD40 in ovarian cancer and adenovirus-mediated CD40 ligand therapy on ovarian cancer in vitro. Tumori, 94(3), 356-361.
[37] Wolf, I., Bose, S., Desmond, J. C., Lin, B. T., Williamson, E. A., Karlan, B. Y., et al. (2007). Unmasking of epigenetically silenced genes reveals DNA promoter methylation and reduced expression of PTCH in breast cancer. Breast Cancer Research and Treatment, 105(2), 139-155.
[38] Yang, Z. Q., Liu, G., Bollig-Fischer, A., Haddad, R., Tarca, A. L., & Ethier, S. P. (2009). Methylation-associated silencing of SFRP1 with an 8p11-12 amplification inhibits canonical and non-canonical WNT pathways in breast cancers. International Journal of Cancer, 125(7), 1613-1621.
[39] Gjerdrum, C., Tiron, C., Hoiby, T., Stefansson, I., Haugen, H., Sandal, T., et al. (2010). Axl is an essential epithelial-to-mesenchymal transition-induced regulator of breast cancer metastasis and patient survival. Proceedings of the National Academy of Sciences of the United States of America, 107(3), 1124-1129.
[40] Miyagi, T., Tatsumi, T., Takehara, T., Kanto, T., Kuzushita, N., Sugimoto, Y., et al. (2003). Impaired expression of proteasome subunits and human leukocyte antigens class I in human colon cancer cells. Journal of Gastroenterology and Hepatology, 18(1), 32-40.
[41] Jang, C. H., Lee, I. A., Ha, Y. R., Lim, J., Sung, M. K., Lee, S. J., et al. (2008). PGK1 induction by a hydrogen peroxide treatment is suppressed by antioxidants in human colon carcinoma cells. Bioscience Biotechnology and Biochemistry, 72(7), 1799-1808.
[42] Tapia-Vieyra, J. V., Arellano, R. O., & Mas-Oliva, J. (2005). ARP2 a novel protein involved in apoptosis of LNCaP cells shares a high degree homology with splicing factor Prp8. Molecular and Cellular Biochemistry, 269(1), 189-201. [43] Coppe, J. P., Itahana, Y., Moore, D. H., Bennington, J. L., & Desprez, P. Y. (2004). Id-1 and Id-2 proteins as molecular markers for human prostate cancer progression. Clinical Cancer Research, 10(6), 2044-2051.
[44] Li, J., Zhang, X. L., Sejas, D. P., Bagby, G. C., & Pang, Q. S. (2004). Hypoxia-induced nucleophosmin protects cell death through inhibition of p53. Journal of Biological Chemistry, 279(40), 41275-41279.
[45] Kokavec, J., Podskocova, J., Zavadil, J., & Stopka, T. (2008). Chromatin remodeling and SWI/SNF2 factors in human disease. Frontiers in Bioscience, 13, 6126-6134.
[46] Schmutzler, C., Mentrup, B., Schomburg, L., Hoang-Vu, C., Herzog, V., & Kohrle, J. (2007). Selenoproteins of the thyroid gland: expression, localization and possible function of glutathione peroxidase 3. Biological Chemistry, 388(10), 1053-1059.
[47] Bundey, R. A., & Insel, P. A. (2006). Adenylyl cyclase 6 overexpression decreases the permeability of endothelial monolayers via preferential enhancement of prostacyclin receptor function. Molecular Pharmacology, 70(5), 1700-1707.
[48] Johansen, L. D., Naumanen, T., Knudsen, A., Westerlund, N., Gromova, I., Junttila, M., et al. (2008). IKAP localizes to membrane ruffles with filamin A and regulates actin cytoskeleton organization and cell migration. Journal of Cell Science, 121(6), 854-864.
[49] Yang, S. C., Cho, Y. S., Chennathukuzhi, V. M., Underkoffler, L. A., Loomes, K., & Hecht, N. B. (2004). Translin-associated factor X is post-transcriptionally regulated by its partner protein TB-RBP, and both are essential for normal cell proliferation. Journal of Biological Chemistry, 279(13), 12605-12614.
[50] Konishi, N., Shimada, K., Nakamura, M., Ishida, E., Ota, I., Tanaka, N., et al. (2008). Function of junB in transient amplifying cell senescence and progression of human prostate cancer. Clinical Cancer Research, 14(14), 4408-4416.
[51] Mullican, S. E., Zhang, S., Konopleva, M., Ruvolo, V., Andreeff, M., Milbrandt, J., et al. (2007). Abrogation of nuclear receptors Nr4a3 and Nr4a1 leads to development of acute myeloid leukemia. Nature Medicine, 13(6), 730-735.
[52] Ganapathi, M., Chikamori, K., Hill, J., Grabowski, D., Zarkhin, E., Grozav, A., et al. (2006). Down regulation of topoisomerase II beta in myeloid leukemia cell lines leads to activation of apoptosis following all-trans retinoic acid-induced differentiation/growth arrest. Ejc Supplements, 4(12), 150-150.
[53] Bansal, M., Belcastro, V., Ambesi-Impiombato, A., & di Bernardo, D. (2007). How to infer gene networks from expression profiles. Molecular Systems Biology, 3(78), 1-10.
[54] Virmani, A. K., Tsou, J. A., Siegmund, K. D., Shen, L. Y. C., Long, T. I., Laird, P. W., et al. (2002). Hierarchical clustering of lung cancer cell lines using DNA methylation markers. Cancer Epidemiology Biomarkers & Prevention, 11(3), 291-297.
[55] Fawcett, T. (2006). An introduction to ROC analysis. Pattern Recognition Letters, 27(8), 861-874.
[56] Chung, M. T., Lai, H. C., Sytwu, H. K., Yan, M. D., Shih, Y. L., Chang, C. C., et al. (2009). SFRP1 and SFRP2 suppress the transformation and invasion abilities of cervical cancer cells through Wnt signal pathway. Gynecologic Oncology, 112(3), 646-653.
[57] Morimoto, K., Kim, S. J., Tanei, T., Shimazu, K., Tanji, Y., Taguchi, T., et al. (2009). Stem cell marker aldehyde dehydrogenase 1-positive breast cancers are characterized by negative estrogen receptor, positive human epidermal growth factor receptor type 2, and high Ki67 expression. Cancer Science, 100(6), 1062-1068.
[58] Oehler, M. K., Fischer, D. C., Orlowska-Volk, M., Herrle, F., Kieback, D. G., Rees, M. C. P., et al. (2003). Tissue and plasma expression of the angiogenic peptide adrenomedullin in breast cancer. British Journal of Cancer, 89(10), 1927-1933.
[59] Toziu-Kara, S., Roux, V., Andrieu, C., Vendrell, J., Vacher, S., Lazar, V., et al. (2007). Oligonucleotide microarray analysis of estrogen receptor alpha-positive postmenopausal breast carcinomas: identification of HRPAP20 and TIMELESS as outstanding candidate markers to predict the response to tamoxifen. Journal of Molecular Endocrinology, 39(3-4), 305-318.
[60] Huhtinen, K., Suvitie, P., Hiissa, J., Junnila, J., Huvila, J., Kujari, H., et al. (2009). Serum HE4 concentration differentiates malignant ovarian tumours from ovarian endometriotic cysts. British Journal of Cancer, 100(8), 1315-1319.
[61] Weaver, A. M., & Silva, C. M. (2006). Modulation of signal transducer and activator of transcription 5b activity in breast cancer cells by mutation of tyrosines within the transactivation domain. Molecular Endocrinology, 20(10), 2392-2405.
[62] Lee, W. R., Chen, C. C., Liu, S. X., & Safe, S. (2006). 17 beta-estradiol (E2) induces cdc25A gene expression in breast cancer cells by genomic and non-genomic pathways. Journal of Cellular Biochemistry, 99(1), 209-220.
[63] Banerjee, A. (2002). Increased levels of tyrosinated alpha-, beta(III)-, and beta(IV)-tubulin isotypes in paclitaxel-resistant MCF-7 breast cancer cells. Biochemical and Biophysical Research Communications, 293(1), 598-601.
[64] Arnold, J. M., Huggard, P. R., Cummings, M., Ramm, G. A., & Chenevix-Trench, G. (2005). Reduced expression of chemokine (C-C motif) ligand-2 (CCL2) in ovarian adenocarcinoma. British Journal of Cancer, 92(11), 2024-2031.
[65] Citterio, C., Vichi, A., Pacheco-Rodriguez, G., Aponte, A. M., Moss, J., & Vaughan, M. (2008). Unfolded protein response and cell death after depletion of brefeldin A-inhibited guanine nucleotide-exchange protein GBF1. Proceedings of the National Academy of Sciences of the United States of America, 105(8), 2877-2882.
[66] Wang, H., Sun, X. Q., Luo, Y., Lin, Z. X., & Wu, J. (2006). Adapter protein NRBP associates with Jab1 and negatively regulates AP-1 activity. Febs Letters, 580(25), 6015-6021.
[67] Mizobe, T., Tsukada, J., Higashi, T., Mouri, F., Matsuura, A., Tanikawa, R., et al. (2007). Constitutive association of MyD88 to IRAK in HTLV-I-transformed T cells. Experimental Hematology, 35(12), 1812-1822.
[68] Varma, D., Dujardin, D. L., Stehman, S. A., & Vallee, R. B. (2006). Role of the kinetochore/cell cycle checkpoint protein ZW10 in interphase cytoplasmic dynein function. Journal of Cell Biology, 172(5), 655-662.