Gen Ekspresyon Verilerinde Yapay Sinir Ağlarına Dayalı Yeni Bir Denetimli Temel Bileşenler Analizi’nin Geliştirilmesi

Bu çalışmada, denetimli temel bileşenler analizi (D-TBA) ile yeni bir yaklaşım olarak önerilen yapay sinir ağlarıyla denetimli temel bileşenler analizi (D-YSA-TBA) kullanılarak çok boyutlu gen ekspresyon verilerinin boyutunun indirgenmesi ve random survival forests (RSF) analizi kullanılarak performansların karşılaştırılması amaçlandı. Simülasyon uygulamasında çok değişkenli normal dağılımdan 100 birim için 5000 gen ve bu gen verisi ile ilişkili yaşam süresi verisi türetildi. Simülasyon aşaması 1000 tekrarlı olarak gerçekleştirildi. Ayrıca yaygın B-hücreli lenfoma (DLBCL) hastası 240 bireye ilişkin gen ekspresyon verileri kullanıldı. Önemli genlerin seçiminde Wald istatistiği kullanılarak boyut indirgemesi yapıldı. Yöntemlerden elde edilen yeni veri setleri RSF analizi kullanılarak analiz edildi. Simülasyon uygulamasında D-TBA ve D-YSA-TBAyöntemlerinin açıklayıcılıkları arasında anlamlı bir fark olduğu görülmüştür (p<0.001). DLBCL verisi ile yapılan uygulamada D-TBA yönteminin hatasının %36.78, D-YSA-TBA yönteminin ise RSF sonucu- %43 olduğu bulunmuştur. D-TBA yönteminin önem değeri diğer yöntemden daha büyük, hatası ise daha düşük çıkmıştır. Çok boyutluluk problemi yaşanan gen ekspresyon verilerinin analizinde D-TBA, D-YSA-TBA’ya göre daha iyi performans göstermiştir.

Anahtar Kelimeler:

Boyut indirgeme; yapay sinir ağları; denetimli temel bileşenler analizi; random survival forests; gen ekspresyon

Development of a New Supervised Principal Component Analysis Based on Artificial Neural Networks in Gene Expression Data

The aim of this study is dimension reduction of multidimensional gene expression data using supervised principal component analysis (S-PCA) and –proposed as a new approach- supervised principal component analysis with artificial neural networks (S-ANN-PCA) and to compare performances of these two methods by using random survival forests (RSF). In simulation application 5000 genes were generated according to multivariate normal distribution and then survival time that is correlated to these gene data were generated for 100 units. Simulation step was carried out with 1000 repetitions. In addition, gene expression data for 240 individuals with extensive B-cell lymphoma (DLBCL) were used. Dimension reduction was done using Wald statistic in selection of important genes. The new data sets obtained from the methods were analyzed using RSF analysis.In the simulation application, it was obtained that the explanatoriness of S-PCA was significantly different from S-ANN-PCA (p<0.001). In the DLBCL data application, it was found that the error rate for the S-PCA was 36.78% and 43% for the S-ANN-PCA as a result of RSF. The importance value of S-PCA method was found to be higher and its error rate was found to be lower than the other method.S-PCA performed better than S-ANN-PCA in analyzing gene expression data experiencing a multidimensional problem.

Keywords:

Dimension reduction; Neural networks; Supervised principal component analysis; Random survival forests; Gene expression,

PDF

___

1. Rosenwald A, Wright G, Chan WC, Connors JM, Campo E, Fisher RI, et al. The use of molecular profiling to predict survival after chemotherapy for diffuse large-B-cell lymphoma. New England Journal of Medicine. 2002;346(25):1937-47.
2. Bair E, Tibshirani R. Semi-supervised methods to predict patient survival from gene expression data. PLoS biology. 2004;2(4):e108.
3. Chen X, Wang L, Smith JD, Zhang B. Supervised principal component analysis for gene set enrichment of microarray data with continuous or survival outcomes. Bioinformatics. 2008;24(21):2474-81.
4. Beer DG, Kardia SL, Huang C-C, Giordano TJ, Levin AM, Misek DE, et al. Gene-expression profiles predict survival of patients with lung adenocarcinoma. Nature medicine. 2002;8(8):816-24.
5. Kramer MA. Nonlinear principal component analysis using autoassociative neural networks. AIChE journal. 1991;37(2):233-43.
6. Hsieh WW. Machine learning methods in the environmental sciences: Neural networks and kernels: Cambridge university press; 2009.
7. Monahan AH. Nonlinear principal component analysis by neural networks: theory and application to the Lorenz system. Journal of Climate. 2000;13(4):821-35.
8. Scholz M, Fraunholz M, Selbig J. Nonlinear principal component analysis: neural network models and applications. Principal manifolds for data visualization and dimension reduction: Springer; 2008. p. 44-67.
9. Dong D, McAvoy TJ. Batch tracking via nonlinear principal component analysis. AIChE Journal. 1996;42(8):2199-208.
10. Hayat EA, Mevlut T, Senol S. An Alternative Dimension Reduction Approach to Supervised Principal Components Analysis in High Dimensional Survival Data. Turkiye Klinikleri Journal of Biostatistics. 2016;8(1):21-9.
11. Albanis G, Batchelor R, editors. Assessing the long-term credit standing using dimensionality reduction techniques based on neural networks—an alternative to overfitting. The proceedings of the SCI 99/ISAS 99 conference, Orlando, US; 1999.
12. Hsieh WW. Nonlinear principal component analysis by neural networks. Tellus A: Dynamic Meteorology and Oceanography. 2001;53(5):599-615.
13. Ture M, Kurt I, Akturk Z. Comparison of dimension reduction methods using patient satisfaction data. Expert Systems with Applications. 2007;32(2):422-6.
14. Oja E. Principal components, minor components, and linear neural networks. Neural networks. 1992;5(6):927-35.
15. Fotheringhame D, Baddeley R. Nonlinear principal components analysis of neuronal spike train data. Biological Cybernetics. 1997;77(4):283-8.
16. Daszykowski M, Walczak B, Massart D. A journey into low-dimensional spaces with autoassociative neural networks. Talanta. 2003;59(6):1095-105.
17. Michailidis G, de Leeuw J. Multilevel homogeneity analysis with differential weighting. Computational statistics & data analysis. 2000;32(3):411-42.
18. Bøvelstad HM, Nygård S, Størvold HL, Aldrin M, Borgan Ø, Frigessi A, et al. Predicting survival from microarray data—a comparative study. Bioinformatics. 2007;23(16):2080-7.
19. Hanley JA, McNeil BJ. A method of comparing the areas under receiver operating characteristic curves derived from the same cases. Radiology. 1983;148(3):839-43.
20. Haykin S. Neural Networks, a comprehensive foundation,2nd ed., Prentice Hall, 842 p. 1999.
21. Breiman L. Random forests. Machine learning. 2001;45(1):5-32.
22. Ishwaran H, Kogalur UB. Random survival forests for R. R News. 2007;7(2):25-31.
23. Nguyen TS, Rojo J. Dimension reduction of microarray data in the presence of a censored survival response: a simulation study. Statistical applications in genetics and molecular biology. 2009;8(1):1-38.
24. Van Wieringen WN, Kun D, Hampel R, Boulesteix A-L. Survival prediction using gene expression data: a review and comparison. Computational statistics & data analysis. 2009;53(5):1590-603.
25. Zhao H, Ljungberg B, Grankvist K, Rasmuson T, Tibshirani R, Brooks JD. Gene expression profiling predicts survival in conventional renal cell carcinoma. PLoS medicine. 2005;3(1):e13.
26. Liu B, Cui Q, Jiang T, Ma S. A combinational feature selection and ensemble neural network method for classification of gene expression data. BMC bioinformatics. 2004;5(1):136.
27. O'Neill MC, Song L. Neural network analysis of lymphoma microarray data: prognosis and diagnosis near-perfect. BMC bioinformatics. 2003;4(1):13.
28. Khan J, Wei JS, Ringner M, Saal LH, Ladanyi M, Westermann F, et al. Classification and diagnostic prediction of cancers using gene expression profiling and artificial neural networks. Nature medicine. 2001;7(6):673-9.
29. Quackenbush J. Computational analysis of microarray data. Nature reviews genetics. 2001;2(6):418-27.
30. Zhang H, Yu C-Y, Singer B, Xiong M. Recursive partitioning for tumor classification with gene expression microarray data. Proceedings of the National Academy of Sciences. 2001;98(12):6730-5.
31. Dudoit S, Fridlyand J, Speed TP. Comparison of discrimination methods for the classification of tumors using gene expression data. Journal of the American statistical association. 2002;97(457):77-87.