论文部分内容阅读
The accumulation of genome-wide gene expression data provide the opportunity to classify cancer into different subgroups based on gene expression signatures.However,due to the heterogeneity of biological data,it is hard to normalize and adjust the batch effects of different datasets.As a reference to human intelligence,we communicate concept or knowledge rather than "data".So,we propose a novel strategy to identify cancer subtypes based on signature-level knowledge network.