Iterative feature removal yields highly discriminative pathways

BMC Genomics

Table 1 Overview of data sets

Data set	Training samples	Testing samples	Features	Classes
Influenza 14-16 [10]	51	57	12023	Asymptomatic, Symptomatic
Influenza 11-14 [10]	68	75	12023	Asymptomatic, Symptomatic
Lung Cancer [11]	32	149	12533	Mesothelioma, Adenocarcinoma
Prostate Cancer [12]	102	34	12600	Tumor, Normal
BCell Lymphoma [13]	47	–	4026	Germinal, Activated

Influenza 14-16 and 11-14 represent different temporal intervals of the data from [10]. The former is used to compare with other published results, and the latter is used because it provides more training samples for the automated analysis. All data sets except the BCell lymphoma have defined training and test partitions.

ISSN: 1471-2164