Correction of gene expression data: performance-dependency on inter-replicate and inter-treatment biases

Research output: Contribution to journal/Conference contribution in journal/Contribution to newspaperJournal articleResearchpeer-review

  • Behrooz Darbani Shirvanehdeh, Denmark
  • C Neal Stewart Jr, Department of Plant Sciences, University of Tennessee-Knoxville, United States
  • Shahin Noeparvar, Denmark
  • Søren Borg
This report investigates for the first time the potential inter-treatment bias source of cell number for gene expression studies. Cell-number bias can affect gene expression analysis when comparing samples with unequal total cellular RNA content or with different RNA extraction efficiencies. For maximal reliability of analysis, therefore, comparisons should be performed at the cellular level. This could be accomplished using an appropriate correction method that can detect and remove the inter-treatment bias for cell-number. Based on inter-treatment variations of reference genes, we introduce an analytical approach to examine the suitability of correction methods by considering the inter-treatment bias as well as the inter-replicate variance, which allows use of the best correction method with minimum residual bias. Analyses of RNA sequencing and microarray data showed that the efficiencies of correction methods are influenced by the inter-treatment bias as well as the inter-replicate variance. Therefore, we recommend inspecting both of the bias sources in order to apply the most efficient correction method. As an alternative correction strategy, sequential application of different correction approaches is also advised.
Original languageEnglish
JournalJournal of Biotechnology
Pages (from-to)100-109
Number of pages9
Publication statusPublished - 20 Oct 2014

    Research areas

  • Data correction, Gene expression, Inter-replicate bias, Inter-treatment bias

See relations at Aarhus University Citationformats

ID: 79725300