Unexpected variability of allelic imbalance estimates from RNA sequencing

RNA sequencing and other experimental methods producing large amounts of data are increasingly dominant in molecular biology. However, the noise properties of these techniques are not fully appreciated. We assessed how reproducible are the measurements of allele-specific expression between replicate RNA-seq experiments from the same RNA sample. Surprisingly, estimates of allelic imbalance (AI) varied between technical replicates up to 8-fold higher than expected from commonly applied noise models. We show that AI overdispersion substantially varies between replicates and experimental series, appears to arise during the construction of sequencing libraries, and can be measured by comparing technical replicates. We demonstrate that compensation for AI overdispersion greatly reduces technical variation and enables reliable differential analysis of allele-specific expression across samples and across experiments. Conversely, not taking AI overdispersion into account can lead to a substantial number of false positives in analysis of allele-specific gene expression. RNA-seq was performed in mouse kidney tissue samples and immortalised B cell line, Abl.1 and Abl.2 clonal cell lines of 129xCastF1 background [1,2]. Libraries (with replicates) were sequenced and data was analyzed. [1] Zwemer, L.M., Zak, A., Thompson, B.R., Kirby, A., Daly, M.J., Chess, A., and Gimelbrant, A.A. (2012). Autosomal monoallelic expression in the mouse. Genome Biol. 13, R10. [2] Nag, A., Savova, V., Fung, H.-L., Miron, A., Yuan, G.-C., Zhang, K., Gimelbrant, A.A., and Gingeras, T. (2013). Chromatin signature of widespread monoallelic expression. eLife 2, e01256.