Poster Annual Conference of the Genetics Society of Australasia with the NZ Society for Biochemistry & Molecular Biology

WITHDRAWN propr: an r package for measuring associations in any -omics data (702)

Thomas Quinn 1 , Mark Richardson 1 , David Lovell 2 , Tamsyn Crowley 1
  1. Deakin University, Waurn Ponds, VIC, Australia
  2. Queensland University of Technology, Brisbane, Queensland, Australia

Correlation is a commonly used method for measuring the association between genomic elements. However, correlation yields spurious (i.e., falsely positive) results when applied to relative data. In contrast to absolute data, relative data contain measurements that only carry meaning when compared to another measurement (e.g., when the two values [50, 100] equate to [500, 1000]). Common examples of relative data include anything measured in percent or parts per million (ppm). However, relative data also can include biological data sets produced by high-throughput RNA-sequencing, chromatin immunoprecipitation (ChIP), ChIP-sequencing, or Methyl-Capture sequencing. Here, we present propr: an R package implementation of proportionality analysis that provides a valid alternative to correlation that is suitable for any and all data sets. Unlike correlation, proportionality yields the same result for relative data as its absolute counter-part, all without generating spurious results. Using a real data set, we show how propr can fit within a larger -omics pipeline to enrich the findings of conventional differential expression analysis. Unlike other analytical pipelines, this one makes no assumption about the underlying distribution of the data and does not require any normalization.