Hello all,

I have been getting stuck on methylation workflow via R while following this guide:

https://nbis-workshop-epigenomics.readthedocs.io/en/latest/content/tutorials/methylationArray/Array_Tutorial.html

This happens near the end, where I am to create a design matrix and contrast matrix. I am running 8 samples, where they are all in separate groups as I am just trying to compare them individually to each other to look for certain methylated or unmethylated genes in the sample.

Nothing would work for the contMatrix, until I set one of the individuals as the 'reference' as shown below.

contMatrix <- makeContrasts(
  VLDFI48_vs_VLDFI44 = individualVLDFI48 - individualVLDFI44,
  VLDFJ48_vs_VLDFI44 = individualVLDFJ48 - individualVLDFI44,
  VLDFJ52_vs_VLDFI44 = individualVLDFJ52 - individualVLDFI44,
  VLDFJ68_vs_VLDFI44 = individualVLDFJ68 - individualVLDFI44,
  VLDFJ97_vs_VLDFI44 = individualVLDFJ97 - individualVLDFI44,
  VLDFK15_vs_VLDFI44 = individualVLDFK15 - individualVLDFI44,
  VLDMT6455_vs_VLDFI44 = individualVLDMT6455 - individualVLDFI44,
  levels = design)

The following error occurs, rightfully so, as there are no replicates:

Error in .ebayes(fit = fit, proportion = proportion, stdev.coef.lim = stdev.coef.lim,  : 
  No residual degrees of freedom in linear model fits

I am supposed to go on to make a variable called fit2 as below:

# fit the contrasts
fit2 <- contrasts.fit(fit, contMatrix)
# Rank genes
fit2 <- eBayes(fit2)

This is needed for annotation, and I tried just using 'fit' that I made a couple lines (to avoid this error) earlier with the error saying I need to run eBayes or treat first.

My question is: am I doing too much? I just want to assess the copy number plot for certain arms of chromosome 1, as well as assess if MGMT along with other genes are methylated or not. Do I have the required data already based off of the guide link I first sent? Everything has already been normalized.

Thank you to anyone that can assist.

Best,
Roy