Identifying differential expression for RNA-seq data with no replication

Transcriptional process is a starting point of biological function. In particular, transcriptomes that display differential expression in different conditions are likely to be key elements in understanding mechanisms causing those differences. A number of statistical approaches have been suggested for discovery of differential expression in microarray platform with replicated samples. However, many of sequencing-based studies tend to have very small or even no replicated sample due to high cost. Because accurate variance estimation is not straightforward with no replicated sample, accurate testing of differential expression is not easy. Here, we propose the permutation-based local pooled error (pLPE) method. By permuting local genes, pLPE method estimates variance more reliably and, thus, facilitates a differential expression analysis with no replicated sample.