Discrepency Between Methylated_Sites.tsv and motifs.tsv

[Ryan-Kich]

I have a question regarding a discrepancy between the methylated_sites.tsv files and the motifs.tsv files. For all of my samples, motifs.tsv reports a .95 to .999 methylation rate for each motif. However, in an independent analysis of pulling all motifs (where the total number of motifs found matches the motifs.tsv output) and aligning all the positions from methylated_sites.tsv, I see proportions ranging from .75 to .90. In addition, the sum of all methylated sites reported in motifs.tsv is consistently quite a bit lower than the sum of rows in my methylated_sites.tsv files. Is there any explanation of why this may be?

I really appreciate your help!

[alexcritschristoph]

Hey @Ryan-Kich ,

That's a good question. I'm thinking of a few possibilities.

Can you report the # of sites and average percentage methylations for:

1. as reported in the motifs file
2. as calculated from the methylated_sites.tsv, for rows with "motif_raw" assigned to the motif_raw.
3. As calculated from the methylated_sites.tsv, for rows with "motif" assigned to the trimmed motif.

I don't necessarily think that'll solve your problem here, because the code is pretty clear, the Average_Percent_Methylation_per_site in the motif table is calculated from the mean of the methylated_sites dataframe: motif_data.Percent_modified.mean().

Could this have to do with multiple methylation types? Are you sure you're calculating methylation of just one type (e.g., just 5mC, not 4mC and 5mC)?

Could it also have to do with coverage? Can you filter the table to min_coverage? I don't think this explains it, but just checking.