1、2018/10/10,DIMACS Workshop on Machine Learning Techniques in Bioinformatics,1,Cancer Classification with Data-dependent Kernels,Anne Ya Zhang (with Xue-wen Chen & Huilin Xiong) EECS & ITTC University of Kansas,2018/10/10,DIMACS Workshop on Machine Learning Techniques in Bioinformatics,2,Outline,Intr
2、oduction Data-dependent Kernel Results Conclusion,2018/10/10,DIMACS Workshop on Machine Learning Techniques in Bioinformatics,3,Cancer facts,Cancer is a group of many related diseases Cells continue to grow and divide and do not die when they should. Changes in the genes that control normal cell gro
3、wth and death. Cancer is the second leading cause of death in the United States Cancer causes 1 of every 4 deaths NIH estimate overall costs for cancer in 2004 at $189.8 billion ($64.9 billion for direct medical cost) Cancer types Breast cancer, Lung cancer, Colon cancer, Death rates vary greatly by
4、 cancer type and stage at diagnosis,2018/10/10,DIMACS Workshop on Machine Learning Techniques in Bioinformatics,4,Motivation,Why do we need to classify cancers? The general way of treating cancer is to: Categorize the cancers in different classes Use specific treatment for each of the classes Tradit
5、ional way to classify cancers Morphological appearanceNot accurate! Enzyme-based histochemical analyses. Immunophenotyping. Cytogenetic analysis.Complicated & needs highly specialized laboratories,2018/10/10,DIMACS Workshop on Machine Learning Techniques in Bioinformatics,5,Motivation,Why traditiona
6、l ways are not enough ? There exists some tumors in the same class with completely different clinical courses May be more accurate classification is needed Assigning new tumors to known cancer classes is not easy e.g. assigning an acute leukemia tumor to one of the AML (acute myeloid leukemia) ALL (
7、acute lymphoblastic leukemia),2018/10/10,DIMACS Workshop on Machine Learning Techniques in Bioinformatics,6,DNA Microarray-based Cancer Diagnosis,Cancer is caused by changes in the genes that control normal cell growth and death. Molecular diagnostics offer the promise of precise, objective, and sys
8、tematic cancer classification These tests are not widely applied because characteristic molecular markers for most solid tumors have to be identified. Recently, microarray tumor gene expression profiles have been used for cancer diagnosis.,2018/10/10,DIMACS Workshop on Machine Learning Techniques in
9、 Bioinformatics,7,Microarray,A microarray experiment monitors the expression levels for thousands of genes simultaneously. Microarray techniques will lead to a more complete understanding of the molecular variations among tumors, hence to a more reliable classification.,2018/10/10,DIMACS Workshop on
10、 Machine Learning Techniques in Bioinformatics,8,Microarray,Microarray analysis allows the monitoring of the activities of thousands of genes over many different conditions. From a machine learning point of view,The large volume of the data requires the computational aid in analyzing the expression
11、data.,2018/10/10,DIMACS Workshop on Machine Learning Techniques in Bioinformatics,9,Machine learning tasks in cancer classification,There are three main types of machine learning problems associated with cancer classification: The identification of new cancer classes using gene expression profiles T
12、he classification of cancer into known classes The identifications of “marker” genes that characterize the different cancer classes In this presentation, we focus on the second type of problems.,2018/10/10,DIMACS Workshop on Machine Learning Techniques in Bioinformatics,10,Project Goals,To develop a
13、 more systematic machine learning approach to cancer classification using microarray gene expression profiles.Use an initial collection of samples belonging to the known classes of cancer to create a “class predictor” for new, unknown, samples.,2018/10/10,DIMACS Workshop on Machine Learning Techniqu
14、es in Bioinformatics,11,Challenges in cancer classification,Gene expression data are typically characterized by high dimensionality (i.e. a large number of genes) small sample sizeCurse of dimensionality!,Methods Kernel techniques Data resampling Gene selection,AML,2018/10/10,DIMACS Workshop on Mach
15、ine Learning Techniques in Bioinformatics,12,Outline,Introduction Data-dependent Kernel Results Conclusion,2018/10/10,DIMACS Workshop on Machine Learning Techniques in Bioinformatics,13,Data-dependent kernel model,Optimizing the data-dependent kernel is to choose the coefficient vector,Data dependen
16、t,2018/10/10,DIMACS Workshop on Machine Learning Techniques in Bioinformatics,14,Optimizing the kernel,Criterion for kernel optimizationMaximum class separability of the training data in the kernel-induced feature space,2018/10/10,DIMACS Workshop on Machine Learning Techniques in Bioinformatics,15,T
17、he Kernel Optimization,In reality, the matrix N0 is usually singular,: eigenvector corresponding to the largest eigenvalue,2018/10/10,DIMACS Workshop on Machine Learning Techniques in Bioinformatics,16,Kernel optimization,Before Kernel Optimization,After Kernel Optimization,Training data,Test data,2
18、018/10/10,DIMACS Workshop on Machine Learning Techniques in Bioinformatics,17,Distributed resampling,Original training data: Training data with resampling:,2018/10/10,DIMACS Workshop on Machine Learning Techniques in Bioinformatics,18,Gene selection,A filter method: class separability,2018/10/10,DIM
19、ACS Workshop on Machine Learning Techniques in Bioinformatics,19,Outline,Introduction Data-dependent Kernel Results Conclusion,2018/10/10,DIMACS Workshop on Machine Learning Techniques in Bioinformatics,20,Comparison with other methods,k-Nearest Neighbor (kNN) Diagonal linear discriminant analysis (
20、DLDA) Uncorrelated Linear Discriminant analysis (ULDA) Support vector machines (SVM),2018/10/10,DIMACS Workshop on Machine Learning Techniques in Bioinformatics,21,Data sets,AML,Subtypes: ALL vs. AML,Status of Estrogen receptor,Status of lymph nodal,Outcome of treatment,Tumor vs. healthy tissue,Subt
21、ypes: MPM vs. ADCA,Different lymphomas cells,Cancer vs. non-cancer,Tumor vs. healthy tissue,2018/10/10,DIMACS Workshop on Machine Learning Techniques in Bioinformatics,22,Experimental setup,Data normalization Zero mean and unity variance at the gene direction Random partition data into two disjoint
22、subsets of equal size training data + test data Repeat each experiment 100 times,2018/10/10,DIMACS Workshop on Machine Learning Techniques in Bioinformatics,23,Parameters,DLDA: no parameter KNN: Euclidean distance, K=3 ULDA: K=3 SVM: Gaussian kernel, use leave-one-out on the training data to tune pa
23、rameters KerNN: Gaussian kernel for basic kernel k0, 0 andare empirically set. Use leave-one-out on the training data to tune the rest parameters. KNN for classification,2018/10/10,DIMACS Workshop on Machine Learning Techniques in Bioinformatics,24,Effect of data resampling,Prostate 102 samples,Lung
24、 181 samples,2018/10/10,DIMACS Workshop on Machine Learning Techniques in Bioinformatics,25,Effect of gene selection,ALL-AML,2018/10/10,DIMACS Workshop on Machine Learning Techniques in Bioinformatics,26,Effect of gene selection,Colon,2018/10/10,DIMACS Workshop on Machine Learning Techniques in Bioi
25、nformatics,27,Effect of gene selection,Prostate,2018/10/10,DIMACS Workshop on Machine Learning Techniques in Bioinformatics,28,Comparison results,ALL-AML,BreastER,BreastLN,Colon,2018/10/10,DIMACS Workshop on Machine Learning Techniques in Bioinformatics,29,Comparison results,CNS,lung,Ovarian,Prostat
26、e,2018/10/10,DIMACS Workshop on Machine Learning Techniques in Bioinformatics,30,Outline,Introduction Data-dependent Kernel Results Conclusion,2018/10/10,DIMACS Workshop on Machine Learning Techniques in Bioinformatics,31,Conclusion,By maximizing the class separability of training data, the data-dep
27、endent kernel is also able to increase the separability of test data. The kernel method is robust to high dimensional microarray data The distributed resampling strategy helps to alleviate the problem of overfitting,2018/10/10,DIMACS Workshop on Machine Learning Techniques in Bioinformatics,32,Concl
28、usion,The classifier assign samples more accurately than other approaches so we can have better treatments respectively. The method can be used for clarifying unusual cases e.g. a patient which was diagnosed as AML but with atypical morphology. The method can be applied to distinctions relating to f
29、uture clinical outcomes.,2018/10/10,DIMACS Workshop on Machine Learning Techniques in Bioinformatics,33,Future work,How to estimate the parameters Study the genes selected,2018/10/10,DIMACS Workshop on Machine Learning Techniques in Bioinformatics,34,Reference,H. Xiong, M.N.S. Swamy, and M.O. Ahmad.
30、 Optimizing the data-dependent kernel in the empirical feature space. IEEE Trans. on Neural Networks 2005, 16:460-474. H. Xiong, Y. Zhang, and X. Chen. Data-dependent Kernels for Cancer Classification. Under review. A. Ben-Dor, L. Bruhn, N. Friedman, I. Nachman, M. Schummer, and Z. Yakhini. Tissue c
31、lassification with gene expression profiles. J. Computational Biology 2000, 7:559-584. S. Dudoit, J. Fridlyand, and T.P. Speed. Comparison of discrimination method for the classification of tumor using gene expression data. J. Am. Statistical Assoc. 2002, 97:77-87 T.S. Furey, N. Cristianini, N. Duff
32、y, D.W. Bednarski, M. Schummer, and D. Haussler. Support vector machine classification and validation of cancer tissue samples using microarray expression data. Bioinformatics 2000, 16:906-914. J. Ye, T. Li, T. Xiong, and R. Janardan. Using uncorrelated discriminant analysis for tissue classification with gene expression data. IEEE/ACM Trans. on Computational Biology and Bioinformatics 2004, 1:181-190.,2018/10/10,DIMACS Workshop on Machine Learning Techniques in Bioinformatics,35,Thanks! Questions?,
