Are there systematic biases in RNA-seq data analysis? A case study for Amphimedon queenslandica sponge as a model object.

BACKGROUNDThe performance of a functional annotation approach for RNA-seq bioinformatics pipelines was to be compared with the method where groups of genes are generated with no relation to ontologes. Three publicly available RNA-Seq experiments for Amphimedon queenslandica sponge were used for the designed comparison. One of these experiments was referred in the publication where stages of embryo development were compared for a wide range of animal species. METHODSThe expression levels were re-calculated here for three independent series of experiments. The functional annotation of differential expression levels was than conducted. This allow to compare an applicability of the two approaches, and to re-evaluate the interpretation provided in the mentioned publication. RESULTSIt was confirmed by the conventional approach that Wnt and Notch pathways do operate in a development of a sponge embryo. The method of annotation which uses unbounded grouping of genes was effective in an ability to separate development stages of sponge embryo. In addition, the published results were by a suggestion distorted by an artifact, caused by a positive feedback in the stage of data processing.