Using normalization to resolve RNA-Seq biases caused by amplification from minimal input

Abstract

RNA-Seq has become a widely used method to study transcriptomes, and it is now possible to perform RNA-Seq on almost any sample. Nevertheless, samples obtained from small cell populations are particularly challenging, as biases associated with low amounts of input RNA can have strong and detrimental effects on downstream analyses. Here we compare different methods to normalize RNA-Seq data obtained from minimal input material. Using RNA from isolated medaka pituitary cells, we have amplified material from six samples before sequencing. Both synthetic and real data are used to evaluate different normalization methods to obtain a robust and reliable pipeline for analysis of RNA-Seq data from samples with very limited input material. The analysis outlined here shows that quantile normalization outperforms other more commonly used normalization procedures when using amplified RNA as input and will benefit researchers employing low amounts of RNA in similar experiments.