Conneely lab software distribution

Annotation and R code for PC-based correction for population stratification in 450K methylation data can be found here.

CpGassoc: an R function for analysis of DNA methylation microarray data

Version 2.11 (for R versions ≥ R 3.0.0)

Version 1.6 (for R versions < R 3.0.0)


Please report errors or suggestions to Richard Barfield at rbarfield01@fas.harvard.edu.

Users who prefer a dropdown menu approach should also check out our GUI software MethLAB, which has many of the same capabilities and can perform global methylation analysis.

To receive emails about future version releases, please email  rbarfield01@fas.harvard.edu  to be added to our mailing list.

For information on installing binary R packages, see http://cran.r-project.org/doc/manuals/R-admin.html#Installing-packages.  To download R, visit http://cran.r-project.org/.  For large datasets (e.g. Illumina 450K data), we recommend running 64-bit R to avoid running into R's memory limits.

Thank you for your interest in CpGassoc!  To reference CpGassoc in a publication, please cite: Barfield RT, Kilaru V, Smith AK, Conneely KN.  CpGassoc: an R Function for analysis of DNA methylation microarray data.  Bioinformatics, 2012; 28(9):1280-1.

Current releases

Version 2.11 (built under R 3.0, compatible with newer versions of R) and Version 1.6 (compatible with older versions of R): 
  1. Fixed bug that was preventing the fdr.method="qvalue" from working in cpg.combine
  2. Fixed bug that was preventing random effects from working when there is missing data in the phenotype 
(Released 1/10/2014)

Older version history

  • Version 1.5/2.0: Fixed minor bugs in manhattan plot function and cpg.qc. Changed the default for QQ plots to not display genomic control factor (lambda) unless requested.  Version 2.0 was built under R 3.0 (released 9/16/2013).
  • Version 1.3: Added genomic control p-values. Added genomic control function cpg.GC. Fixed bug in manhattan plot function and plot function with mixed effect model t-statistics. Fixed bug in subset option of cpg.assoc.
  • Version 1.2: Improved finding of sites with extreme missingness that corresponded to the main phenotype of interest Version 1.1: Fixed bug in the function 'cpg.qc'.
  • Version 1.00: Can now save files as .eps and .pdf. Can specify titles for plots. Added genomic inflation factor based on the p-values for the qq-plots. Fixed a minor problem with performing permutation tests with mixed effects models for when a model fails to converge in a permutation. Slight changes in options for plot functions. See help pages for more information.
  • Version 0.95: Fixed small bug in scatterplot function. Version 0.9: Added a quality control function (cpg.qc) for Illumina data. Updated Help Pages Version 0.8: manhattan plot now allows users to set color scheme, only plot a certain region, or label the significant sites.  Also now includes a tutorial in the doc section of the package.
  • Version 0.7: Circumvented problem with phenotypes of very small variance.  Version 0.6: Fixed problem with scatterplot function. Also corrected some small problems with a categorical phenotype Version 0.5: Nicer Manhattan plots. User can now specify the covariates as a formula. Added further corrections for  possibility of user having a covariate of one unique value.  Version 0.4: Fixed bug in the function 'design' that would have affected categorical analysis. Also added std. error to  standard output (for tests of continuous independent variables only).
  • Version 0.3: Fixed bug in the function 'design'. 
  • Version 0.2: Beta values now are required to be one column per individual, one row per CpG site.