Figure 4
From: The identification of informative genes from multiple datasets with increasing complexity
The improvement or deterioration of genes ranking in Sartorelli. Firstly, we selected 100 informative and 50 uninformative genes using Tomczak dataset and extracted their ranks in Sartorelli. Secondly, we trained 1PB classifier on Tomczak and tested on Sartorelli. Finally, we ranked genes according to the average error rate of 1PB classifier in predicting their values in Sartorelli. This figure illustrates the average improvement or deterioration of Myogenesis-Related, Top 100, and 50 randomly selected genes in Sartorelli generated with our method and the gene rankings generated by concordance model.