What's the norm in normalization? A frightening note on the use of RT-qPCR in the livestock science.

Reverse-Transcription quantitative PCR (RT-qPCR) provides a valuable tool to study gene expression with exquisite sensitivity. To retain its inferential power, user-introduced technical variability must be reduced and accounted for. Selecting a set of stably expressed internal control genes (ICG), validated for each experimental condition/sample set, is widely accepted as a reliable way to normalize RT-qPCR data and account for said variability. Despite significant efforts in establishing standardized and resource-efficient normalization approaches, numerous recent reports have underlined deficiencies in the state of RT-qPCR normalization. Livestock science has benefitted tremendously from the use of RT-qPCR; however, the issue of lack of proper normalization likely affects this discipline as well. We thus decided to determine whether this is true, and to which extent. We conducted an in-depth analysis of all (225) RT-qPCR articles published in the six most prominent livestock journals in the field from 2013 to 2017. A quantitative scale was constructed, and values were assigned to each article based on the number of ICG used, the use of a publicly available algorithm to assess the reliability of ICG, and the reporting of pertinent information related to ICG (ranges from 0 = total noncompliance - to 100 = total compliance). Out of the surveyed group, only 10.7% of the publications obtained a score of 100, while the largest group (n = 158) was represented by articles that scored 0. Subdividing articles based on whether an algorithm to validate ICG was used (YAL) or not (NAL) revealed the use of a larger number of ICG to normalize RT-qPCR in the YAL group compared to NAL (1.4-fold more, 95% C.I.: 1.11-1.84) and was closer to the "gold standard" of three ICG. Using an algorithm also increased the diversity of ICG and significantly reduced the use of RNA18S, whose suitability as ICG has been thoroughly debated. These remarkably low normalization standards are likely to generate questionable results that can severely hinder the advance of transcriptomic studies in livestock science and related fields.