Classification of Heterogeneous Microarray Data by Maximum Entropy Kernel Method, Wataru Fujibuchi* and Tsuyoshi Kato* (accepted in BMC Bioinformatics)
*Both authors equally contributed to this work.

References

Kato, T., Fujibuchi, W., and Asai, K. (Jun., 2006) Kernel analysis for noisy microarray data. AIST Technical Report, AIST02-J00001-8 (http://www.cb.k.u-tokyo.ac.jp/asailab/kato/pdf/t-kato-cbrctr2006a.pdf).
Kato, T., Fujibuchi, W., and Asai, K. (Jul., 2006) Learning kernels from distance constraints. IPSJ Transactions on Computer Vision and Image Media, 47(10).

Supplemental files for microarray prediction analysis

1. Renal clear cell carcinoma

all data(renal_carcinoma.xls)

original GEO site([GPL9, GPL10],[GPL91],[GPL96, GPL97],[GPL91, GPL92, GPL93, GPL94, GPL95],[GPL96, GPL1074],[GPL91])

2. AML and ALL data

all data(AML-ALL.xls)

original Golub's data(Broad Institute)

3. Oral cavity carcinoma metastasis

each data(oral_each.xls)

original GEO site(GSE2280, GSE3524)

mixed data(oral_mixed.xls)

top 1000 feature genes in each Leave-One-Out-Cross-Validation in mixed data in tab limited text(oral_mixed_feature.tar.gz)

Kernel Entropy program is available from this page.

*Note: our program uses an interor point algorithm for SVM optimization, which requires a threshold for judging if a training sample is used as support vector. We have tested 1e-4 and 1e-5 for the threshold in our paper, which can be specified by "svm.thres-C" option.