Background

After comparing analyses and looking at genome feature locations, we decided to continue using the Ronit dataset with Yaamini controls to see what the processes look like.

GOterm annotation

GOterm annotation was carried out in a Jupyter notebook (not archived in the project repository).

The key outputs I got from this notebook are:

  1. A master annotation table (cgigas_uk_roslin_v1_rna_from_genomic_annot.transcript.tab).
  2. A summary (union, 1x coverage) of the CpG background generated form the .bed files, joined with transcript IDs and GO annotations (union_1x.GeneIDs.geneOverlap.transcriptIDs.GOAnnot).
  3. A gene ID to GO term map that compares the unique transcript IDs within the background and the blast annotation (geneid2go.tab)
  4. A summary object of the DML associated with the test factor, joined with transcript IDs and GO annotations (e.g., DML-Cov10-20.GeneIDs.geneOverlap.transcriptIDs.GOAnnot)

I created a unique notebook for each dataset. Ronit’s data had a lot of GO terms, and required at least 48 GB of ram to complete the analysis.

Results

Here is the prior table for reference: Table 1. DML counts from each analysis Table of differentially methylated locus counts from each analysis

Figure 1. GO terms (counts + percentages). BP = biological processes, CC = cellular components, MF = molecular functions. Table of enriched GO term counts and percentages by category

Figure 2. Unique transcripts, genes, and gene products (counts). BP = biological processes, CC = cellular components, MF = molecular functions. Table of unique transcripts, genes, and gene products by GO category

Figure 3. Unique DML in Genes w/ Enriched GO terms (counts). Table of unique DML in genes with enriched GO terms (counts)

Figure 4. Unique DML in Genes w/ Enriched GO terms (percentages). Table of unique DML in genes with enriched GO terms (percentages)