This git repository consists of more detailed case reports and simulations made in conjunction with, but not part of, the article "Methods that remove batch effects while retaining group differences may lead to exaggerated confidence in downstream analyses", Vegard Nygaard, Einar Andreas Rødland, and Eivind Hovig, Biostat first published online August 13, 2015, doi:10.1093/biostatistics/kxv027
The target audience is the especially interested ones. The documents are not peer-reviewed and might be changed or updated after the publication data.
The format of the case-reports is html, but the underlying r-code and text is available and rerunnable. Run knit2html or the Rstudio knitr integration on the individual .rmd files to re-produce the plots and html.
Reanalysis and alternative analysis of "Data set 2" from the original ComBat article: Johnson, WE, Rabinovic, A, and Li, C (2007). Adjusting batch effects in microarray expression data using Empirical Bayes methods. Biostatistics 8(1):118-127.
Reanalysis and alternative analysis of parts of Towfic et al.(2014, January). Comparing the biological impact of glatiramer acetate with the biological impact of a generic. PloS one 9(1), e83757.
Full random number re-run of the tutorial from the sva package (v3.8.0), Leek JT, Johnson WE, Parker HS, Jaffe AE, Storey JD.(2012) The sva package for removing batch effects and other unwanted variation in high-throughput experiments. Bioinformatics. 2012 Mar 15;28(6):882-3.
Source code for an interactive web application which demonstrates the effects of batch adjustment using simulated data. Made in the shiny-framework in R and accessible at https://vegard.shinyapps.io/batch-adjustment-simulator.
Source code for an interactive R/shiny application for computing the corrected F-distribution for any group-batch design. Compares this to the F-distribution assumed by one-way ANOVA. Also runs analyses on simulated data to demonstrate the difference. Available at https://vegard.shinyapps.io/fdist-app.
R-script implementing the computations described in the supplementary material section 1 on simulated data.
- helperfunctions.r ad hoc methods.
The scripts that produces the figures in "Methods that remove batch effects while retaining group differences may lead to exaggerated confidence in downstream analyses" are found here: https://github.com/ous-uio-bioinfo-core/batch-adjust-warning-figures.git