Revisiting DMR identification

After talking to Neel, we decided to go back to BAT_filter_vcf step and see if changing some of the filtering settings would lead to DMR identification. By imposing more lenient restrictions at the filtering step, I may have more loci available for metilene to construct methylated regions.

Checking code

Before I proceeded with re-filtering my called samples, I checked the BAT_summarize code to make sure I didn’t mix any samples up! I reviewed the code and fortunately (or unfortunately?) there were no obvious mistakes. I also reviewed the pdf output from BAT_overview to see if the samples sequenced on L4 clustered together. Since I noticed they had more truncated patterns than the other samples, I wanted to make sure the truncation wasn’t impacting my ability to call DMR. I didn’t see any L4 samples cluster together in the dendograms.

Refiltering samples

Alright, now that I checked everything else, it was time to re-filter samples. I tried the following settings:

• MDP_min = 10, MDP_max = 100
• MDP_min = 10
• MDP_min = 10, MDP_max = 500
• MDP_min = 10, MDP_max = 1000

Once samples were re-filtered, I ran them through BAT_summarizeand BAT_DMRcalling. I used four contrasts for these tests: 20 vs. 5 and 20 vs. OC within each population. For DMR_calling, I used a minimum of 1 CpG per DMR just to see if I could get any methylated region detected first, then changed those settings if I identified DMR.

Table 1. Number of DMR identified with different filtering and DMR identification settings

Contrast	Group 1	Group 2	MDP_min = 10, MDP_max = 100, MR_min = .10, c = 10	MDP_min = 10, MDP_max = 100, MR_min = .10, c = 1	MDP_min = 10, c = 1	MDP_min = 10, c = 5	MDP_min = 10, c = 10	MDP_min = 10, MDP_max = 100, c = 1	MDP_min = 10, MDP_max = 500, c = 1	MDP_min = 10, MDP_max = 500, c = 5	MDP_min = 10, MDP_max = 500, c = 10
N	20	5	0	0	16	16	16	0	16	16	16
N	20	OC	0	0	2	2	2	0	2	2	2
S	20	5	0	0	1	1	1	0	1	1	1
s	20	OC	0	0	1	1	1	0	1	1	1

Reviewing my DMR output, it’s clear that the same DMR were called regardless of the settings! Now that I know there are DMR in these datasets, I want to finalize my BAT_filter_vcf and BAT_DMRcalling settings so I can apply them to the other four contrasts.

Going forward

1. Finalize BAT_filter_vcf and BAT_DMRcalling settings
2. Run all data through BAT_filter_vcf, BAT_summarize, BAT_overview, and BAT_DMRcalling
3. Update methods and results
4. Obtain other important methylation landscape information
5. Generate genome feature tracks
6. Annotate DMR locations
7. Try DMR identification with bismark and methylKit
8. Start RNA-Seq analysis
9. Create OSF repository for all intermediate files