A hybrid method for differentially expressed genes identification and ranking from RNA-Seq data

Abstract

RNA-Seq has gained immense popularity and emerged as a potential high-throughput platform for identification of differentially expressed (DE) genes. In order to estimate the nature of differential genes, it is important to find statistical distributional property of the data. In the present study we propose a new hybrid model (NBPFCROS) based on parametric and non-parametric statistic for the identification of DE genes. The NBP model based on Compound mixture of Poisson-gamma distribution is used as a parametric statistic and Fold change value derived using fold change rank ordering statistics (FCROS) algorithm is used as non-parametric statistic, we used a gene significance score pi-value by combining expression fold change (f value) and statistical significance (p-value). The performance of NBPFCROS model was compared with NBP, FCROS, edgeR and DESeq2 models using synthetic and real RNA-Seq datasets and it was found that the developed model NBPFCROS is more robust as compared to the other models.