未來展望

第五章實驗資料與結果討論

6.2 未來展望

最近幾年來蛋白激酶磷酸化的問題已是疾病治療中最有潛力的新興研究領域[29]。

像是酪氨酸激脢接受體 (receptor tyrosine kinase, RTK) 是細胞表面生長因子的一部份，

它有者固有的配體控制的酪氨酸激脢 (tyrosine-kinase) 活性，同時在正常細胞內廣泛調

控許多功能，並且是致癌基因的關鍵。在本研究中我們也預測了三個酪氨酸激酶INSR，

EGFR 跟 SRC。其中表皮生長因子受體(EGFR)的治療，是近年來出現確有療效的抗腫瘤治療。正常細胞的酪胺酸激酶活性一般會受到嚴密的調控，但是癌細胞往往具有很強的酪胺酸激酶活性，使得癌細胞內的酪胺酸激酶接受體過度表現，於是癌細胞不斷的進行分化、增殖、抗凋亡(anti-apoptosis)、血管新生以及轉移，因此酪胺酸激酶被認為和癌

細胞的增生有關。因此若能設法抑制酪胺酸激酶的活性，使酪胺酸激酶接受體不再過度表現，將有助於癌細胞的控制。在使用人化抗體和小分子藥物等干預療法中 RTKs 和生長因子這些配體已經變成合理的標靶，而以 RTK 為基礎的癌症治療，像是轉移性乳癌、胃腸道間質瘤和非小細胞肺癌等等，已經廣泛的應用在臨床上，並且藉此開起了基因治療研究發展的新動力[10]。另外週期素依賴性激脢 (cyclin-dependent kinases, CDK) 當作標靶已經成為[30]中風的診斷與治療策略，以及蛋白激酶抑制劑用來治療心臟衰竭症[31]，另外像是 p38 MAP 激酶也被認為是炎症性疾病治療的主要標靶[32]，這些是本研究未來要發展的方向。

參考文獻

[1] N. Blom, S. Gammeltoft, and S. Brunak, "Sequence and structure-based prediction of eukaryotic protein phosphorylation sites," Journal of Molecular Biology, vol. 294, pp.

1351-62, Dec 17 1999.

[2] R. Linding, L. J. Jensen, G. J. Ostheimer, M. A. van Vugt, C. Jorgensen, I. M. Miron, F.

Diella, K. Colwill, L. Taylor, K. Elder, P. Metalnikov, V. Nguyen, A. Pasculescu, J. Jin, J. G. Park, L. D. Samson, J. R. Woodgett, R. B. Russell, P. Bork, M. B. Yaffe, and T.

Pawson, "Systematic discovery of in vivo phosphorylation networks," Cell, vol. 129, pp.

1415-26, Jun 29 2007.

[3] H. D. Huang, T. Y. Lee, S. W. Tzeng, L. C. Wu, J. T. Horng, A. P. Tsou, and K. T.

Huang, "Incorporating hidden Markov models for identifying protein kinase-specific phosphorylation sites," Journal of Computational Chemistry, vol. 26, pp. 1032-41, Jul 30 2005.

[4] S. R. Eddy, "Profile hidden Markov models," Bioinformatics, vol. 14, pp. 755-763, 1998.

[5] F. Diella, S. Cameron, C. Gemund, R. Linding, A. Via, B. Kuster, T. Sicheritz-Ponten, N. Blom, and T. J. Gibson, "Phospho.ELM: a database of experimentally verified phosphorylation sites in eukaryotic proteins," BMC Bioinformatics, vol. 5, p. 79, Jun 22 2004.

[6] M. Huse and J. Kuriyan, "The conformational plasticity of protein kinases," Cell, vol.

109, pp. 275-282, May 3 2002.

[7] S. K. Hanks, A. M. Quinn, and T. Hunter, "The protein kinase family: conserved features and deduced phylogeny of the catalytic domains," Science, vol. 241, pp. 42-52, Jul 1 1988.

[8] S. K. Hanks and T. Hunter, "Protein kinases 6. The eukaryotic protein kinase superfamily: kinase (catalytic) domain structure and classification," FASEBJ Journal, vol. 9, pp. 576-96, May 1995.

[9] G. Manning, D. B. Whyte, R. Martinez, T. Hunter, and S. Sudarsanam, "The protein kinase complement of the human genome," Science, vol. 298, pp. 1912-34, Dec 6 2002.

[10] A. Gschwind, O. M. Fischer, and A. Ullrich, "Timeline - The discovery of receptor tyrosine kinases: targets for cancer therapy," Nature Reviews Cancer, vol. 4, pp.

361-370, May 2004.

[11] D. R. Knighton, J. H. Zheng, L. F. Teneyck, V. A. Ashford, N. H. Xuong, S. S. Taylor, and J. M. Sowadski, "Crystal-Structure of the Catalytic Subunit of Cyclic Adenosine-Monophosphate Dependent Protein-Kinase," Science, vol. 253, pp. 407-414, Jul 26 1991.

[12] G. E. Crooks, G. Hon, J. M. Chandonia, and S. E. Brenner, "WebLogo: A sequence logo generator," Genome Research, vol. 14, pp. 1188-1190, Jun 2004.

[13] T. D. Schneider and R. M. Stephens, "Sequence Logos - a New Way to Display Consensus Sequences," Nucleic Acids Research, vol. 18, pp. 6097-6100, Oct 25 1990.

[14] L. R. Rabiner, "A Tutorial on Hidden Markov-Models and Selected Applications in Speech Recognition," Proceedings of the IEEE, vol. 77, pp. 257-286, Feb 1989.

[15] A. Krogh, M. Brown, I. S. Mian, K. Sjolander, and D. Haussler, "Hidden Markov-Models in Computational Biology - Applications to Protein Modeling,"

Journal of Molecular Biology, vol. 235, pp. 1501-1531, Feb 4 1994.

[16] G. A. Churchill, "Stochastic models for heterogeneous DNA sequences," Bulletin of Mathematical Biology, vol. 51, pp. 79-94, 1989.

[17] J. F. Mari, J. P. Haton, and A. Kriouile, "Automatic word recognition based on second-order hidden Markov models," IEEE Transactions on Speech and Audio Processing, vol. 5, pp. 22-25, Jan 1997.

[18] I. Shahin, "Improving speaker identification performance under the shouted talking condition using the second-order hidden Markov models," Eurasip Journal on Applied Signal Processing, vol. 2005, pp. 482-486, Mar 15 2005.

[19] I. Shahin, "Enhancing speaker identification performance under the shouted talking condition using second-order circular hidden Markov models," Speech Communication, vol. 48, pp. 1047-1055, Aug 2006.

[20] O. Aycard, J.-F. Mari, and R. Washington, "Learning to automatically detect features for mobile robots using second-order Hidden Markov Models," International Journal Of Advanced Robotic Systems vol. 1, pp. 231-244, Dec 2004.

[21] A. Krogh and G. Mitchison, "Maximum entropy weighting of aligned sequences of proteins or DNA," Proceedings Third International Conference on Intelligent Systems for Molecular Biology vol. 3, pp. 215-21, 1995.

[22] K. Sjolander, K. Karplus, M. Brown, R. Hughey, A. Krogh, I. S. Mian, and D. Haussler,

"Dirichlet mixtures: A method for improved detection of weak but significant protein sequence homology," Computer Applications in the Biosciences, vol. 12, pp. 327-345, Aug 1996.

[23] H. Akaike, "New Look at Statistical-Model Identification," IEEE Transactions on Automatic Control, vol. Ac19, pp. 716-723, 1974.

[24] G. Schwarz, "Estimating Dimension of a Model," Annals of Statistics, vol. 6, pp.

461-464, 1978.

[25] A. Schliep, I. G. Costa, C. Steinhoff, and A. Schonhuth, "Analyzing gene expression time-courses," IEEE/ACM Transactions on Computational Biology and Bioinformatics, vol. 2, pp. 179-93, Jul-Sep 2005.

[26] J. Martin, J. F. Gibrat, and F. Rodolphe, "Analysis of an optimal hidden Markov model for secondary structure prediction," BMC Structural Biology, vol. 6, p. 25, 2006.

[27] M. Brand, "Structure learning in conditional probability models via an entropic prior and parameter extinction," Neural Computation, vol. 11, pp. 1155-1182, Jul 1 1999.

[28] U. J. Matthew A. Siegler, Bhiksha Raj, Richard M. Stern, "Automatic Segmentation, Classification and Clustering of Broadcast News Audio," DARPA Speech Recognition Workshop, 1997.

[29] R. Sridhar, O. Hanson-Painton, and D. R. Cooper, "Protein kinases as therapeutic targets," Pharmaceutical Research, vol. 17, pp. 1345-1353, Nov 2000.

[30] H. Osuga, S. Osuga, F. H. Wang, R. Fetni, M. J. Hogan, R. S. Slack, A. M. Hakim, J. E.

Ikeda, and D. S. Park, "Cyclin-dependent kinases as a therapeutic target for stroke,"

Proceedings of the National Academy of Sciences of the United States of America, vol.

97, pp. 10254-10259, Aug 29 2000.

[31] C. J. Vlahos, S. A. McDowell, and A. Clerk, "Kinases as therapeutic targets for heart failure," Nature Reviews Drug Discovery, vol. 2, pp. 99-113, Feb 2003.

[32] S. Kumar, J. Boehm, and J. C. Lee, "p38 map kinases: Key signalling molecules as therapeutic targets for inflammatory diseases," Nature Reviews Drug Discovery, vol. 2, pp. 717-726, Sep 2003.

附錄: 詳細實驗數據

PKA (S)

全名：Protein kinase A

資料筆數：positive 跟 negative 各 308 筆資料序列長度：15

磷酸化位置：中間的絲氨酸(S)

圖 A.1: PKA (S)資料的序列圖案

表A.1: PKA (S)的 30 次 5-CV 於測試資料的效能比較表

PKA (S) Threshold Sensitivity Specificity Precision Accuracy HMMer -4.2667

(0.2835)

38 iHMM - PKA (S)

0.4 0.5 0.6 0.7 0.8 0.9 1

-10 -8.1 -6.2 -4.3 -2.4 -0.5 1.4 3.3 5.2 7.1 9 10.9 12.8 14.7

Threshold

Accuracy

Training set Test set

圖 A.2: PKA (S)資料於 iHMM 的門檻值與正確率對應圖

HMMer - PKA (S)

0.4 0.5 0.6 0.7 0.8 0.9 1

-25 -22 -20 -17 -14 -12 -9.1 -6.5 -3.8 -1.2 1.5 4.15 6.8 9.45

Threshold

Accuracy

Training set Test set

圖 A.3: PKA (S)資料於 HMMer 的門檻值與正確率對應圖

ROC of training set of PKA (S)

0 0.2 0.4 0.6 0.8 1

1-Specificity

Sensitivity

iHMM HMMer

圖 A.4: PKA (S)於訓練資料的 ROC 圖

ROC of test set of PKA (S)

0 0.2 0.4 0.6 0.8 1

1-Specificity

Sensitivity

iHMM HMMer

圖 A.5: PKA (S)於測試資料的 ROC 圖

40 PKA (S)

0.4 0.5 0.6 0.7 0.8 0.9 1

Training set accuracy

Test set accuracy

iHMM HMMer iHMM correlation coefficient = 0.9990

HMMer correlation coefficient = 0.9980

圖 A.6: PKA (S)資料的相關係數分析圖

PKA (T)

全名：Protein kinase A

資料筆數：positive 跟 negative 各 39 筆資料序列長度：15

磷酸化位置：中間的蘇氨酸(T)

圖 A.7: PKA (T)資料的序列圖案

表A.2: PKA (T)的 30 次 5-CV 於測試資料的效能比較表

PKA (T) Threshold Sensitivity Specificity Precision Accuracy HMMer ^-2.3400 (28.0175)

0.7430

42 iHMM - PKA (T)

0.4 0.5 0.6 0.7 0.8 0.9 1

-15 -13 -10 -8.1 -5.8 -3.5 -1.2 1.1 3.4 5.7 8 10.3 12.6 14.9

Threshold

Accuracy

Training set Test set

圖 A.8: PKA (T)資料於 iHMM 的門檻值與正確率對應圖

HMMer - PKA(T)

0.4 0.5 0.6 0.7 0.8 0.9 1

-30 -27 -23 -20 -17 -13 -9.9 -6.5 -3.2 0.15 3.5 6.85 10.2 13.6

Threshold

Accuracy

Training set Test set

圖 A.9: PKA (T)資料於 HMMer 的門檻值與正確率對應圖

ROC of training set of PKA (T)

0 0.2 0.4 0.6 0.8 1

1-Specificity

Sensitivity

iHMM HMMer

圖 A.10: PKA (T)於訓練資料的 ROC 圖

ROC of test set of PKA (T)

0 0.2 0.4 0.6 0.8 1

1-Specificity

Sensitivity

iHMM HMMer

圖 A.11: PKA (T)於測試資料的 ROC 圖

44 PKA (T)

0.4 0.5 0.6 0.7 0.8 0.9 1

Training set accuracy

Test set accuracy

iHMM HMMer iHMM correlation coefficient = 0.9972

HMMer correlation coefficient = 0.6182

圖 A.12: PKA (T)資料的相關係數分析圖

PKB (S)

全名：Protein kinase B

資料筆數：positive 跟 negative 各 81 筆資料序列長度：15

磷酸化位置：中間的絲氨酸(S)

圖 A.13: PKB (S)資料的序列圖案

表A.3: PKB (S)的 30 次 5-CV 於測試資料的效能比較表

PKB (S) Threshold Sensitivity Specificity Precision Accuracy HMMer ^-5.4843

46 iHMM - PKB (S)

0.4 0.5 0.6 0.7 0.8 0.9 1

-15 -13 -11 -8.3 -6 -3.8 -1.5 0.75 3 5.25 7.5 9.75 12 14.3

Threshold

Accuracy

Training set Test set

圖 A.14: PKB (S)資料於 iHMM 的門檻值與正確率對應圖

HMMer - PKB (S)

0.4 0.5 0.6 0.7 0.8 0.9 1

-30 -27 -23 -20 -17 -13 -9.9 -6.5 -3.2 0.15 3.5 6.85 10.2 13.6

Threshold

Accuracy

Training set Test set

圖 A.15: PKB (S)資料於 HMMer 的門檻值與正確率對應圖

ROC of training set of PKB (S)

0 0.2 0.4 0.6 0.8 1

1-Specificity

Sensitivity

iHMM HMMer

圖 A.16: PKB (S)於訓練資料的 ROC 圖

ROC of test set of PKB (S)

0 0.2 0.4 0.6 0.8 1

1-Specificity

Sensitivity

iHMM HMMer

圖 A.17: PKB (S)於測試資料的 ROC 圖

48 PKB (S)

0.4 0.5 0.6 0.7 0.8 0.9 1

Training set accuracy

Test set accuracy

iHMM HMMer iHMM correlation coefficient = 0.9962

HMMer correlation coefficient = 0.9280

圖 A.18: PKB (S)資料的相關係數分析圖

PKC (S)

全名：Protein kinase C

資料筆數：positive 跟 negative 各 304 筆資料序列長度：15

磷酸化位置：中間的絲氨酸(S)

圖 A.19: PKC (S)資料的序列圖案

表A.4: PKC (S)的 30 次 5-CV 於測試資料的效能比較表

PKC (S) Threshold Sensitivity Specificity Precision Accuracy HMMer ^-4.9333

50 iHMM - PKC (S)

0.4 0.5 0.6 0.7 0.8 0.9 1

-15 -13 -11 -8.3 -6 -3.8 -1.5 0.75 3 5.25 7.5 9.75 12 14.3

Threshold

Accuracy

Training set Test set

圖 A.20: PKC (S)資料於 iHMM 的門檻值與正確率對應圖

HMMer - PKC (S)

0.4 0.5 0.6 0.7 0.8 0.9 1

-25 -23 -20 -18 -15 -13 -10 -7.5 -5 -2.5 0 2.5 5 7.5 10

Threshold

Accuracy Training set

Test set

圖 A.21: PKC (S)資料於 HMMer 的門檻值與正確率對應圖

ROC of training set of PKC (S)

0 0.2 0.4 0.6 0.8 1

1-Specificity

Sensitivity

iHMM HMMer

圖 A.22: PKC (S)於訓練資料的 ROC 圖

ROC of test set of PKC (S)

0 0.2 0.4 0.6 0.8 1

1-Specificity

Sensitivity

iHMM HMMer

圖 A.23: PKC (S)於測試資料的 ROC 圖

52 PKC (S)

0.4 0.5 0.6 0.7 0.8 0.9 1

Training set accuracy

Test set accuracy

iHMM HMMer iHMM correlation coefficient = 0.9986

HMMer correlation coefficient = 0.9965

圖 A.24: PKC (S)資料的相關係數分析圖

PKC (T)

全名：Protein kinase C

資料筆數：positive 跟 negative 各 71 筆資料序列長度：15

磷酸化位置：中間的蘇氨酸(T)

圖 A.25: PKC (T)資料的序列圖案

表A.5: PKC (T)的 30 次 5-CV 於測試資料的效能比較表

PKC (T) Threshold Sensitivity Specificity Precision Accuracy HMMer ^-4.3507

54 iHMM - PKC (T)

0.4 0.5 0.6 0.7 0.8 0.9 1

-15 -13 -11 -8.3 -6 -3.8 -1.5 0.75 3 5.25 7.5 9.75 12 14.3

Threshold

Accuracy

Training set Test set

圖 A.26: PKC (T)資料於 iHMM 的門檻值與正確率對應圖

HMMer - PKC (T)

0.4 0.5 0.6 0.7 0.8 0.9 1

-30 -27 -23 -20 -17 -13 -9.9 -6.5 -3.2 0.15 3.5 6.85 10.2 13.6

Threshold

Accuracy

Training set Test set

圖 A.27: PKC (T)資料於 HMMer 的門檻值與正確率對應圖

ROC of training set of PKC (T)

0 0.2 0.4 0.6 0.8 1

1-Specificity

Sensitivity

iHMM HMMer

圖 A.28: PKC (T)於訓練資料的 ROC 圖

ROC of test set of PKC (T)

0 0.2 0.4 0.6 0.8 1

1-Specificity

Sensitivity

iHMM HMMer

圖 A.29: PKC (T)於測試資料的 ROC 圖

56 PKC (T)

0.4 0.5 0.6 0.7 0.8 0.9 1

Training set accuracy

Test set accuracy

iHMM HMMer iHMM correlation coefficient = 0.9582

HMMer correlation coefficient = 0.9386

圖 A.30: PKC (T)資料的相關係數分析圖

PKG (S)

全名：Protein kinase G

資料筆數：positive 跟 negative 各 30 筆資料序列長度：15

磷酸化位置：中間的絲氨酸(S)

圖 A.31: PKG (S)資料的序列圖案

表A.6: PKG (S)的 30 次 5-CV 於測試資料的效能比較表

PKG (S) Threshold Sensitivity Specificity Precision Accuracy HMMer ^-4.9190

(0.4089) HMM-1 ^-0.31667

(0.1448)

iHMM ^-0.31667 (0.1448) HMMer ^-13.0930

(1.4398) 0.9030

(0.0542) 0.8055

(0.0752) 0.8551

(0.0489) 0.8658 (0.0243) HMM-1 ^1.4277

(1.3280) 0.7972

(0.1074) 0.7314

(0.0952) 0.7560

(0.0845) 0.7894 (0.0241) δ2

iHMM ^1.4277 (1.3280)

58 iHMM - PKG (S)

0.4 0.5 0.6 0.7 0.8 0.9 1

-15 -13 -11 -8.3 -6 -3.8 -1.5 0.75 3 5.25 7.5 9.75 12 14.3

Threshold

Accuracy

Training set Test set

圖 A.32: PKG (S)資料於 iHMM 的門檻值與正確率對應圖

HMMer - PKG (S)

0.4 0.5 0.6 0.7 0.8 0.9 1

-30 -27 -23 -20 -17 -13 -9.9 -6.5 -3.2 0.15 3.5 6.85 10.2 13.6

Threshold

Accuracy

Training set Test set

圖 A.33: PKG (S)資料於 HMMer 的門檻值與正確率對應圖

ROC of training set of PKG (S)

0 0.2 0.4 0.6 0.8 1

1-Specificity

Sensitivity

iHMM HMMer

圖 A.34: PKG (S)於訓練資料的 ROC 圖

ROC of test set of PKG (S)

0 0.2 0.4 0.6 0.8 1

1-Specificity

Sensitivity

iHMM HMMer

圖 A.35: PKG (S)於測試資料的 ROC 圖

60 PKG (S)

0.4 0.5 0.6 0.7 0.8 0.9 1

Training set accuracy

Test set accuracy

iHMM HMMer iHMM correlation coefficient = 0.9381

HMMer correlation coefficient = 0.6653

圖 A.36: PKG (S)資料的相關係數分析圖

CDK (S)

全名：Cyclin-dependent kinase

資料筆數：positive 跟 negative 各 195 筆資料序列長度：15

磷酸化位置：中間的絲氨酸(S)

圖 A.37: CDK (S)資料的序列圖案

表A.7: CDK (S)的 30 次 5-CV 於測試資料的效能比較表

CDK (S) Threshold Sensitivity Specificity Precision Accuracy HMMer ^-4.0093

(0.2652) HMM-1 ^1.8995

(0.2686)

iHMM ^3.3438 (0.2076) HMMer ^-5.0437

(0.4899) 0.8456

(0.0232) 0.8558

(0.0214) 0.8588

(0.0152) 0.8507 (0.0073) HMM-1 ^1.9771

(0.3029) 0.8326

(0.0227) 0.8361

(0.0190) 0.8387

(0.0142) 0.8344 (0.0090) δ2

iHMM ^3.6848 (0.2914)

62 iHMM - CDK (S)

0.4 0.5 0.6 0.7 0.8 0.9 1

-15 -13 -10 -7.5 -5 -2.5 0 2.5 5 7.5 10 12.5

Threshold

Accuracy

Training set Test set

圖 A.38: CDK (S)資料於 iHMM 的門檻值與正確率對應圖

HMMer - CDK (S)

0.4 0.5 0.6 0.7 0.8 0.9 1

-30 -26 -23 -19 -15 -12 -7.8 -4.1 -0.4 3.3 7

Threshold

Accuracy

Training set Test set

圖 A.39: CDK (S)資料於 HMMer 的門檻值與正確率對應圖

ROC of training set of CDK (S)

0 0.2 0.4 0.6 0.8 1

1-Specificity

Sensitivity

iHMM HMMer

圖 A.40: CDK (S)於訓練資料的 ROC 圖

ROC of test set of CDK (S)

0 0.2 0.4 0.6 0.8 1

1-Specificity

Sensitivity

iHMM HMMer

圖 A.41: CDK (S)於測試資料的 ROC 圖

64 CDK (S)

0.4 0.5 0.6 0.7 0.8 0.9 1

Training set accuracy

Test set accuracy

iHMM HMMer iHMM correlation coefficient = 0.9970

HMMer correlation coefficient = 0.9961

圖 A.42: CDK (S)資料的相關係數分析圖

CDK (T)

全名：Cyclin-dependent kinase

資料筆數：positive 跟 negative 各 113 筆資料序列長度：15

磷酸化位置：中間的蘇氨酸(T)

圖 A.43: CDK (T)資料的序列圖案

表A.8: CDK (T)的 30 次 5-CV 於測試資料的效能比較表

CDK (T) Threshold Sensitivity Specificity Precision Accuracy HMMer ^-4.6577

66 iHMM - CDK (T)

0.4 0.5 0.6 0.7 0.8 0.9 1

-15 -13 -11 -8.3 -6 -3.8 -1.5 0.75 3 5.25 7.5 9.75 12 14.3

Threshold

Accuracy Training set

Test set

圖 A.44: CDK (T)資料於 iHMM 的門檻值與正確率對應圖

HMMer - CDK (T)

0.4 0.5 0.6 0.7 0.8 0.9 1

-30 -27 -23 -20 -17 -13 -9.9 -6.5 -3.2 0.15 3.5 6.85 10.2 13.6

Threshold

Accuracy

Training set Test set

圖 A.45: CDK (T)資料於 HMMer 的門檻值與正確率對應圖

ROC of training set of CDK (T)

0 0.2 0.4 0.6 0.8 1

1-Specificity

Sensitivity

iHMM HMMer

圖 A.46: CDK (T)於訓練資料的 ROC 圖

ROC of test set of CDK (T)

0 0.2 0.4 0.6 0.8 1

1-Specificity

Sensitivity

iHMM HMMer

圖 A.47: CDK (T)於測試資料的 ROC 圖

68 CDK (T)

0.4 0.5 0.6 0.7 0.8 0.9 1

Training set accuracy

Test set accuracy

iHMM HMMer iHMM correlation coefficient = 0.9970

HMMer correlation coefficient = 0.9474

圖 A.48: CDK (T)資料的相關係數分析圖

CaM-KII (S)

全名：Calcium/calmodulin-dependent protein kinase II 資料筆數：positive 跟 negative 各 76 筆資料

序列長度：15

磷酸化位置：中間的絲氨酸(S)

圖 A.49: CaM-KII (S)資料的序列圖案

表A.9: CaM-KII (S)的 30 次 5-CV 於測試資料的效能比較表

CaM-KII (S) Threshold Sensitivity Specificity Precision Accuracy HMMer ^-4.4003

iHMM - CaM-KII (S)

0.4 0.5 0.6 0.7 0.8 0.9 1

-12 -10 -8.7 -7 -5.4 -3.8 -2.1 -0.5 1.2 2.8 4.5 6.1 7.8 9.4

Threshold

Accuracy Training set

Test set

圖 A.50: CaM-KII (S)於 iHMM 的門檻值與正確率對應圖

HMMer - CaM-KII (S)

0.4 0.5 0.6 0.7 0.8 0.9 1

-30 -27 -23 -20 -17 -13 -9.9 -6.5 -3.2 0.15 3.5 6.85 10.2 13.6

Threshold

Accuracy

Training set Test set

圖 A.51: CaM-KII (S)於 HMMer 的門檻值與正確率對應圖

ROC of training set of CaM-KII (S)

0 0.2 0.4 0.6 0.8 1

1-Specificity

Sensitivity

iHMM HMMer

圖 A.52: CaM-KII (S)於訓練資料的 ROC 圖

ROC of test set of CaM-KII (S)

0 0.2 0.4 0.6 0.8 1

1-Specificity

Sensitivity

iHMM HMMer

圖 A.53: CaM-KII (S)於測試資料的 ROC 圖

72 CaM-KII (S)

0.4 0.5 0.6 0.7 0.8 0.9 1

Training set accuracy

Test set accuracy

iHMM HMMer iHMM correlation coefficient = 0.9908

HMMer correlation coefficient = 0.9596

圖 A.54: CaM-KII (S)資料的相關係數分析圖

CK1 (S) 全名：Casein kinase I

資料筆數：positive 跟 negative 各 49 筆資料序列長度：15

磷酸化位置：中間的絲氨酸(S)

圖 A.55: CK1 (S)資料的序列圖案

表A.10: CK1 (S)的 30 次 5-CV 於測試資料的效能比較表

CK1 (S) Threshold Sensitivity Specificity Precision Accuracy HMMer ^-4.6300

74 iHMM - CK1 (S)

0.4 0.5 0.6 0.7 0.8 0.9 1

-10 -8.1 -6.2 -4.3 -2.4 -0.5 1.4 3.3 5.2 7.1 9 10.9 12.8 14.7

Threshold

Accuracy

Training set Test set

圖 A.56: CK1 (S)資料於 iHMM 的門檻值與正確率對應圖

HMMer - CK1 (S)

0.4 0.5 0.6 0.7 0.8 0.9 1

-30 -27 -23 -20 -17 -13 -9.9 -6.5 -3.2 0.15 3.5 6.85

Threshold

Accuracy

Training set Test set

圖 A.57: CK1 (S)資料於 HMMer 的門檻值與正確率對應圖

ROC of training set of CK1 (S)

0 0.2 0.4 0.6 0.8 1

1-Specificity

Sensitivity

iHMM HMMer

圖 A.58: CK1 (S)於訓練資料的 ROC 圖

ROC of test set of CK1 (S)

0 0.2 0.4 0.6 0.8 1

1-Specificity

Sensitivity

iHMM HMMer

圖 A.59: CK1 (S)於測試資料的 ROC 圖

76 CK1 (S)

0.4 0.5 0.6 0.7 0.8 0.9 1

Training set accuracy

Test set accuracy

iHMM HMMer iHMM correlation coefficient = 0.9924

HMMer correlation coefficient = 0.6927

圖 A.60: CK1 (S)資料的相關係數分析圖

CK2 (S)

全名：Casein kinase II

資料筆數：positive 跟 negative 各 243 筆資料序列長度：15

磷酸化位置：中間的絲氨酸(S)

圖 A.61: CK2 (S)資料的序列圖案

表A.11: CK2 (S)的 30 次 5-CV 於測試資料的效能比較表

CK2 (S) Threshold Sensitivity Specificity Precision Accuracy

在文檔中以二階隱藏式馬可夫模型預測特定蛋白激酶磷酸化的位置 (頁 45-122)

第五章 實驗資料與結果討論

6.2 未來展望

參考文獻

附錄: 詳細實驗數據

第五章實驗資料與結果討論