How Radmeth merge define the beggining and the end of a DMR

[Quetzely Ortiz]

Dear methpipe group,

I am searching for DMRs between Arabidopsis samples. I ran the radmeth pipeline to compare groups of methylomes. I am wondering how radmeth merge define the beginning and the end of a DMR? According to the MethPipe software manual, it joins neighboring differentially methylated sites with p-value below 0.01 however in my methylome analysis some DMRs contain 1 significantly differentially methylated CpGs but include more than 10 CpGs in that region. I suspect some DMRs contain CpGs with p-value above the cutoff value (0.01), in that case how much CpGs with p-value above the cutoff value are allowed insed each DMR. 

Thank you so much. The best,

Quetzely.

[Xiaojing Ji]

Hi Quetzely,

Not sure if the forwarded conversation answers your question. It would also be helpful if you can provide us the sample data: one example DMR and the radmeth results of included CpGs.

Thank you,

Liz

--
You received this message because you are subscribed to the Google Groups "MethPipe and MethBase Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to methpipe+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/methpipe/ed677d03-c3dc-4aab-b157-d9291f9e301co%40googlegroups.com.

[Quetzely Ortiz]

Thank you for your quick reply Liz,

I can not see a forwarded conversation. I provide you one example of CG DMR (located in Chr1 coord. 6502    6747) and the radmeth results obtained before the DMR. That DMRs contain 1 significantly differentially methylated CpGs but include 14 CpGs in that region, then how rarmeth merge defined that DMR?.

The output of “radmeth merge -p 0.01 Col0_Vs_mut_CG.adjusted.bed > Col0_Vs_mut_CG.dmrs.bed”

Chr1    6502    6747    dmr     1       0.560144

Chr1    30879   30881   dmr     1       -0.48951

Chr1    31970   32139   dmr     4       -0.0852883

Chr1    42546   42692   dmr     2       -0.569343

 

The output of “radmeth adjust -bins 1:200:1 Col0_Vs_mut_CG.bed > Col0_Vs_mut_CG.adjusted.bed”

Chr1	6502	+	CG	0.273908	1.71E-05	0.00091844	6	5	39	24
Chr1	6503	-	CG	0.119205	1.71E-05	0.00091844	9	8	36	23
Chr1	6512	+	CG	0.0197222	1.71E-05	0.00091844	9	8	38	14
Chr1	6513	-	CG	0.3477	1.71E-05	0.00091844	8	5	34	15
Chr1	6520	+	CG	0.00997149	1.71E-05	0.00091844	8	7	37	1
Chr1	6521	-	CG	0.0379556	1.71E-05	0.00091844	7	3	35	2
Chr1	6683	+	CG	0.0135042	6.13E-06	0.00044019	15	14	35	0
Chr1	6684	-	CG	0.0135009	6.13E-06	0.00044019	16	15	30	0
Chr1	6686	+	CG	0.013019	6.13E-06	0.00044019	17	16	35	0
Chr1	6687	-	CG	0.0117528	6.13E-06	0.00044019	16	11	31	0
Chr1	6692	+	CG	0.0114537	6.13E-06	0.00044019	17	15	30	0
Chr1	6693	-	CG	0.0123026	6.13E-06	0.00044019	15	10	32	0
Chr1	6746	+	CG	0.28336	8.26E-05	0.0028373	16	0	49	2
Chr1	6747	-	CG	-1	-1	-2	17	0	33	0
Chr1	6750	+	CG	0.0766408	8.26E-05	0.0028373	16	0	46	5

 

radmeth regression -factor case Col0_Vs_mut_CG.design.txt  Col0_Vs_mut_CG.proportion.txt

Chr1	6502	+	CG	0.273908	6	5	39	24
Chr1	6503	-	CG	0.119205	9	8	36	23
Chr1	6512	+	CG	0.0197222	9	8	38	14
Chr1	6513	-	CG	0.3477	8	5	34	15
Chr1	6520	+	CG	0.00997149	8	7	37	1
Chr1	6521	-	CG	0.0379556	7	3	35	2
Chr1	6683	+	CG	0.0135042	15	14	35	0
Chr1	6684	-	CG	0.0135009	16	15	30	0
Chr1	6686	+	CG	0.013019	17	16	35	0
Chr1	6687	-	CG	0.0117528	16	11	31	0
Chr1	6692	+	CG	0.0114537	17	15	30	0
Chr1	6693	-	CG	0.0123026	15	10	32	0
Chr1	6746	+	CG	0.28336	16	0	49	2
Chr1	6747	-	CG	-1	17	0	33	0

Thank you very much,

Quetzely

El lunes, 13 de julio de 2020, 21:51:35 (UTC-5), Xiaojing Ji escribió:

Hi Quetzely,

Not sure if the forwarded conversation answers your question. It would also be helpful if you can provide us the sample data: one example DMR and the radmeth results of included CpGs.

Thank you,

Liz

On Jul 13, 2020, at 3:21 PM, Quetzely Ortiz <coba...@gmail.com> wrote:

Dear methpipe group,

I am searching for DMRs between Arabidopsis samples. I ran the radmeth pipeline to compare groups of methylomes. I am wondering how radmeth merge define the beginning and the end of a DMR? According to the MethPipe software manual, it joins neighboring differentially methylated sites with p-value below 0.01 however in my methylome analysis some DMRs contain 1 significantly differentially methylated CpGs but include more than 10 CpGs in that region. I suspect some DMRs contain CpGs with p-value above the cutoff value (0.01), in that case how much CpGs with p-value above the cutoff value are allowed insed each DMR. 

Thank you so much. The best,

Quetzely.

--
You received this message because you are subscribed to the Google Groups "MethPipe and MethBase Users" group.

To unsubscribe from this group and stop receiving emails from it, send an email to meth...@googlegroups.com.

[Xiaojing Ji]

Hi Quetzely,

"radmeth merge" joins CpGs based on their adjusted P-values (combined from neighboring sites), which can be much lower than their original p-values.  However,  "radmeth merge" also keeps track of the number of significantly differentially-methylated CpGs (before adjusting) and reports the number in the 5-th column of the output file. That's why the number may not be equal to the number of CpGs that the DMR spans. 

Best Regards,

Liz

To unsubscribe from this group and stop receiving emails from it, send an email to methpipe+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/methpipe/3a764004-3ccf-4e9b-a474-91b1921041b4o%40googlegroups.com.

[Quetzely Ortiz]

Hi Liz,

Thank you for answering my question. Some DMRs output of Radmeth span more CpGs than the number of CpGs significantly differentially-methylated. In Arabidopsis methylomes after calling DMRs with radmeth merge, do you suggest filter by significant CpGs sites (5-th column of the output file)? ¿Is a Radmeth DMR more relevant with a higher number of significant CpGs sites? or what is the reason that the 5-th column is reported?

I  apologize if the question sounds too basic, I'm a newbie to bioinformatics and don't have too much knowledge on DNA methylation, except some basic idea.

The best,

Quetzely.

To view this discussion on the web visit https://groups.google.com/d/msgid/methpipe/3a764004-3ccf-4e9b-a474-91b1921041b4o%40googlegroups.com.

[Xiaojing Ji]

Hi Quetzely,

Thanks for your question! RADMeth considers the correlation between individual sites to help us identify the start and end of a DMR, so it doesn't simply join differentially-methylated CpGs together. The number and p-values of differentially methylated CpGs do contribute to the combined p-value before merging. The regions reported by "RADMeth merge" must contain at least one significantly differentially methylated site. In general, filtering is not required for finding meaningful DMRs. But whether and how to filter the regions also depend on the specific research problem.

Best,

Liz

To unsubscribe from this group and stop receiving emails from it, send an email to methpipe+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/methpipe/d54c263c-0fa3-4dfc-a21f-5782c1a0f345o%40googlegroups.com.