Identification of cancer subtypes is critically important for understanding the heterogeneity present in tumors. Integrating information from multiple sources, homogeneous groups for cancer can be identified. However, there is a lack of computational approaches to identify histological subtypes among the patients suffering from different types of cancers. Assigning weight to the biomarkers prior to the integration of multiple information sources for the same set of samples can play an important role in cancer subtypes identification, which has not been explored previously. Sub-typing of cancers can help in analyzing shared molecular profiles between different histological subtypes of solid tumors. A novel method for feature weighting based on robust regression fit is developed in this study. The weight is utilized to find similarity between patients individually from each of the information sources. Here, miRNA and mRNA expression profiles across the same set of samples have been used. Patient-similarity networks, that are generated from each of the expression profiles are then integrated using the approach of Similarity Network Fusion. Finally, Spectral clustering is applied on the fused network to identify similar groups of patients that represent a cancer subtype. The effectiveness of the proposed method has been demonstrated on different data sets. IEEE