Explore counts comparisons with an interactive RShiny app.
To run the app from terminal you will need R installed on your device. Use the command R -e "shiny::runApp('shinyApp') where 'shinyApp' is replaced by the path to the directory that app.R was downloaded to on your device.
Alternatively, the app can be run from RStudio by opening app.R in RStudio and clicking on the Run App button in the top right.
The app currently supports input data to be in a SummarizedExperiment Object or count matrix and metadata table pair. For a brief guide on how to format your data as a SummarizedExperiment Object, see format.R. To upload with a count matrix, a metadata table will need to be uploaded as well. For an example of the required metadata table format, see mock_metadata.csv.
PEM Calculator allows users to calculate Preferential Expression Measure (PEM) values from count matrices and explore the analysis results. PEM measures the expression of a gene in a given group within a condition in relation to its expression in all groups within that condition. Therefore, PEM analysis is a useful tool for visualizing a one-to-many comparison of gene expression between groups of a sample condition. PEM analysis provides a score for each gene per group within a given condition. It is recommended that low quality samples and features be removed prior to PEM analysis. Additionally, it is recommended that counts be normalized prior to PEM analysis.
A PEM score is calculated by dividing the observed expression count by an expected expression count and log-10 transforming this value. The observed expression count is the average expression count of a feature in a given group of a condition. The expected expression count in PEM score calculations is the sum of expression values observed for a gene multiplied by the sum of expression values observed for a given group divided by the sum of all expression values observed. A positive PEM score indicates over-expression of a gene and a negative PEM score indicates under-expression of a gene.
A convenience script to install all required packages for the app is provided by install_packages.R.
The PEM metric was initially designed as a tissue-level expression specificity analysis strategy (Huminiecki et al., 2003). A benchmark of tissue expression specificity metrics identified PEM to be the best performing method for estimating the expression specificity of a gene per each tissue analyzed (Kryuchkova-Mostacci & Robinson-Rechavi, 2016). The function used in this app was derived from the supplementary materials of this benchmarking experiment.
Huminiecki, L., Lloyd, A. T., & Wolfe, K. H. (2003). Congruence of tissue expression profiles from Gene Expression Atlas, SAGEmap and TissueInfo databases. BMC Genomics, 4(1), 31. https://doi.org/10.1186/1471-2164-4-31 Kryuchkova-Mostacci, N., & Robinson-Rechavi, M. (2016). A benchmark of gene expression tissue-specificity metrics. Briefings in Bioinformatics, bbw008. https://doi.org/10.1093/bib/bbw008