Visualize Cv DMLs and DMRs

[sr320]

In IGV visualize DMLs and DMRs with corresponding filtered methylation data to verify DMLs / DMRs

[yaaminiv]

To clarify: all DMLs and DMRs?

[sr320]

Yes from your gonad analysis with methylation of all samples thanks Steven

…

On Mar 4, 2019, 2:16 AM -0500, Yaamini Venkataraman ***@***.***>, wrote: To clarify: all DMLs and DMRs? — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or mute the thread.

[yaaminiv]

What criteria should I use to verify DML/DMR? I'll get this done once I'm able to access my bedGraphs from gannet.

[sr320]

We should look at and see how / if it makes sense- not sure bedgraphs will work- pretty sure they are 1x - will need to run some simple code on .cov files. Also have track that highlights common loci would be useful thanks Steven

…

On Mar 4, 2019, 5:05 PM -0500, Yaamini Venkataraman ***@***.***>, wrote: What criteria should I use to verify DML/DMR? I'll get this done once I'm able to access my bedGraphs from gannet. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or mute the thread.

[sr320]

Reminder all files are available as backup on google drive thanks Steven

…

On Mar 4, 2019, 5:05 PM -0500, Yaamini Venkataraman ***@***.***>, wrote: What criteria should I use to verify DML/DMR? I'll get this done once I'm able to access my bedGraphs from gannet. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or mute the thread.

[yaaminiv]

Here's my IGV session. Coverage files (3x and 10x) can be found in this folder. BEDfiles for DML, DMR (100 bp and 1000 bp windows), and the gene background/unite output are in this folder.

[yaaminiv]

Changed paths from local to URL paths. IGV session can be found here. Just looking at heatmaps for DMRs, it seems like they're often artifacts of areas with no data.

[sr320]

genome location is still local - see first few lines of xml file.

[yaaminiv]

genome location is now a URL from gannet.

[sr320]

The "gene background / unite output" is not what you need.

What you want is a track that indicates what CpG are being considered for DML / DMR analysis. This would be all CpGs with 3x coverage in all samples?

[sr320]

I am going to suggest there are some errors in the DMR code (or track development)
Is not the basis that all CpG need 3x coverage for consideration?

[sr320]

More likely a visualization issue rather than data but CpG methylation is offset (bar at left).

[yaaminiv]

What you want is a track that indicates what CpG are being considered for DML / DMR analysis. This would be all CpGs with 3x coverage in all samples?

From what I understand, the unite output is all CpGs with 3x coverage in all samples (from Bioconductor:

The unite() function will return a methylBase object which will be our main object for all comparative analysis. The methylBase object contains methylation information for regions/bases that are covered in all samples.

I specified 3x coverage using mincov with processBismarkAln (step before unite)

[yaaminiv]

I am going to suggest there are some errors in the DMR code (or track development)
Is not the basis that all CpG need 3x coverage for consideration?

Tiles were created using all CpGs with 3x coverage (R Markdown file here):

tiles100 <- tileMethylCounts(processedFilesCov3, win.size = 100, step.size = 100) #Add up C and T counts from each covered cytosine and return total C and T count for each tile. Perform this analysis with 100 bp sized windows, and slide 100 bp for each new window

Not entirely sure what went wrong there...