Differentially methylated genes

[Lei Jiang]

Hi, Altuna,

I have used  methylkit and genomation for DMR analysis and it works very well. I have a question, to identify DMR, I used tileMethylcounts function with specified windowsize and stepsize to define the regions. Then I used the annotateWithGeneParts and getAssociationWithTSS to get their genomic feature to see where the DMR overlaps with specific genes. I am wondering whether it is possible to directly compare the methylation difference on a per-gene basis without defining the window of regions but using the gene length as the window? Is it possible to do that using methylkit and genomation?

Thanks in advance!

Lei

[David Fisher]

Hi Lei,

Yes this is possible. If you have your genes as a GRanges object (this is how I did it, you might have another approach):

ann_granges = read_gff3("../../path/annotation.gff3")

genes_granges = ann_granges %>% filter(type == "gene") #using dplyr functions

you can then use the "regionCounts" function to summarise the methylated and unmethylated Cs for each gene:

#Assume "db" is you methyl database object

counts_gene = regionCounts(db, genes_granges)

You can then use the functions like normal:

lapply(counts_gene, getMethylationStats,  plot=F)
lapply(counts_gene, getCoverageStats,  plot=F)

counts_gene_f = filterByCoverage(counts_gene, hi.count = 100000)  #need to check the plots/info to work out what this should be
counts_gene_fu = methylKit::unite(counts_gene_f, destrand=FALSE, min.per.group = 3L, suffix = "genes_fu3") #some of these options are specifc to what I was doing.

... etc

Hope that helps.

David

[Lei Jiang]

Hi, David,

Thanks so much for the hint! Using read.gff and dplyr will generate a transcript/gene table but not the grange file that can be processed in the following regioncounts analysis. However, I found the following code that would work to generate a grange from gff or gtf file:   makeGRangesFromGFF( file, level = c("genes", "transcripts"), ignoreTxVersion = TRUE, .checkAgainstTxDb = FALSE ) or makeGRangesFromGTF( file, level = c("genes", "transcripts"), ignoreTxVersion = TRUE, .checkAgainstTxDb = FALSE )

Best,

Lei

[Lei Jiang]

Hi everyone!

granges object of gene bodies can also be prepared from the gff file using the gffToGRanges function in the genomation package (Akalin et al. 2014)

[Georges Kanaan]

Hi Lei,

I was wondering if you could share the code that ended up working for you when using gffToGRanges?