A miRNA- and mRNA-seq-Based Feature Selection Approach for Kidney Cancer Biomakers

Microarray data sets have been used for predicting cancer biomarkers. Yet, replication of the prediction has not been fully satisfied. Recently, new data sets called deep sequencing data sets have been generated, with an advantage of less noise in computational analysis. In this study, we analyzed the kidney miRNA and mRNA sequence data sets for predicting cancer markers using 5 different statistical feature selection methods. In the results, we obtained 3 mRNA- and 27 miRNA-based cancer biomarkers to compare with the normal samples. In addition, we clustered the kidney cancer subtypes using a nonnegative matrix factorization method and obtained significant results of survival analysis from the 2 separate groups including miRNA-342 and its target eukaryotic translation initiation factor 5A (EIF5A).