Compositional Data Analysis is necessary for simulating and analyzing RNA-Seq data

*Seq techniques (e.g. RNA-Seq) generate compositional datasets, i.e. the number of fragments sequenced is not proportional to the total RNA present. Thus, datasets carry only relative information, even though absolute RNA copy numbers are often of interest. Current normalization methods assume most features are not changing, which can lead to misleading conclusions when there are large shifts. However, there are few real datasets and no simulation protocols currently available that can directly benchmark methods when such large shifts occur.We present absSimSeq, an R package that simulates compositional data in the form of RNA-Seq reads. We tested several tools used for RNA-Seq differential analysis: sleuth, DESeq2, edgeR, limma, sleuth and ALDEx2 (which explicitly takes a compositional approach). For these tools, we compared their standard normalization to either “compositional normalization”, which uses log-ratios to anchor the data on a set of negative control features, or RUVSeq, another tool that directly uses negative control features.We show that common normalizations result in reduced performance with current methods when there is a large change in the total RNA per cell. Performance improves when spike-ins are included and used by a compositional approach, even if the spike-ins have substantial variation. In contrast, RUVSeq, which normalizes count data rather than compositional data, has poor performance. Further, we show that previous criticisms of spike-ins did not take into account the compositional nature of the data. We conclude that absSimSeq can generate more representative datasets for testing performance, and that spike-ins should be more broadly used in a compositional manner to minimize misleading conclusions from differential analyses.