search

hputnam / Meth_Compare

5 stars 2 forks source link

Confirm # of common 5x CpG loci #87

Closed shellywanamaker closed 4 years ago

shellywanamaker commented 4 years ago

I made this script (https://github.com/hputnam/Meth_Compare/blob/master/analyses/Coverage_analysis/PCA_5xUnionBedCpGs.Rmd) to create PCA plots for common 5x CpG loci from 5x union bedgraphs (generated by 10-Mcap-Canonical-Coverage-Track-STRIGG.ipynb and 11-Pact-Canonical-Coverage-Track-STRIGG.ipynb and located here: https://gannet.fish.washington.edu/metacarcinus/FROGER_meth_compare/20200424/10-unionbedg/)

The number of loci I found covered by all samples at 5x coverage was:

This is different than the number of common loci that @mgavery found in her corr plot analysis. She found:

@mgavery can you confirm that you are filtering for 5x coverage? I found that the MethylKit methRead command by default uses a 10x cutoff in the user manual (https://www.bioconductor.org/packages/release/bioc/vignettes/methylKit/inst/doc/methylKit.html) unless the mincov parameter is specified. Could this be the reason for the discrepancy?

mgavery commented 4 years ago

Removing the filter step and just uniting the samples based on the 5x 'mincov' in methRead

-4666 for Mcap -93714 for Pact

Whew! Thanks SO much Shelly for looking into this!!