Identification of cancer subtypes based on gene expression data plays an important role to develope an appropriate therapy for the patient. However, the analysis of gene expression data related to cancer subtypes identification have many difficulties such as, high dimensional attributes, missing values and sparse data problem. To solve this problem, Iterative Scalling Fuzzy Additive Spectral Clustering (is-FADDIS) will be introduced. This research aims to compare the performance of is-FADDIS to other popular clustering techniques including Gaussian Mixture Clustering and Auto K-Means on the basis of Subtypess Cancer Identification in Human Colorectal Carcinoma and B-Cell Lymphoma dataset. The result of the experiment shows that is-FADDIS successfully produce three cluster structures in Human Colorectal Carcinoma and two well separated cluster structures in B-Cell Lymphoma dataset.