A Method Of Peeling Off The Hidden Genetic Heterogeneities Of Cancers Based On Pathway

Discovering the subtypes of cancer in molecular level is of critical importance both for understanding pathogenic mechanisms and for finding efficient treatments. Recently, it has been recognized the hidden genetic heterogeneities of cancers are related to pathway. So the genes highly variably expressed across the disease samples that are significantly enrich in the same pathway should help to unravel the hidden subtypes of cancer. They were finally used to discover the subtypes of cancer by clustering algorithm. As for validation, we used a large cancer dataset to evaluate the ability of the method of peeling off the hidden genetic heterogeneities of cancers based on pathway for correctly portioning samples. Then, we applied the proposed method to two publicly available microarray datasets of diffuse large B-cell lymphoma (DLBCL), a notoriously heterogeneous phenotype. Finally, the clinical significance of the identified subtypes was verified by survival analysis. In the validation dataset, we achieved highly accurate partitions that best fit the clinical cancer phenotypes. Then, for the two notoriously heterogeneous DLBCL, we had the excellent results. One of them demonstrate that two partitioned subtypes using the functional gene groups had very different 5-year overall rates (62%, 24%) and were highly significantly ( p =0.024) correlated with the clinical survival rate, and the other demonstrate that three partitioned subtypes had very different 10-year overall rates (90%, 46%, 20%) and were highly significantly ( p =0.008) correlated with the clinical survival rate. The proposed approach is a promising computational strategy for peeling off genetic heterogeneities and understanding the modular mechanisms of cancers.