Source:

@ARTICLE{Bittner2000,

author = {M. Bittner and P. Meltzer and Y. Chen and Y. Jiang and E. Seftor and M. Hendrix and M. Radmacher and R. Simon and Z. Yakhini and A. Ben-Dor and N. Sampas and E. Dougherty and E. Wang and F. Marincola and C. Gooden and J. Lueders and A. Glatfelter and P. Pollock and J. Carpten and E. Gillanders and D. Leja and K. Dietrich and C. Beaudry and M. Berens and D. Alberts and V. Sondak},

title = {Molecular classification of cutaneous malignant melanoma by gene expression profiling.},

journal = {Nature},

year = {2000},

volume = {406},

pages = {536--540},

number = {6795},

month = {Aug},

doi = {10.1038/35020115},

url = {http://dx.doi.org/10.1038/35020115}

}

Original data: http://research.nhgri.nih.gov/microarray/Melanoma_Supplement/Files/gene_list-Cutaneous_Melanoma.xls

Description:

The authors report the discovery of a subset of melanomas identified by mathematical analysis of gene expression in a series of samples. In order to so, they gathered expression profile for 38 samples, including 31 melanomas and 7 controls. The resulting hierarchical clustering of the 31 melanoma samples demonstrates that 19 samples are tightly clustered. We denote the class of these samples by ML1. The others, including the 7 controls, will be referred to as ML2. That is, the samples are distributed within the classes as follows:

• Bittner: 19 ML1 and 19 ML2.

Parameters used in our filter:

• Bittner: l= 4 and c=29