Hello,
I am using modkit dmr and would like clarification on the interpretation of the cohen_h_l and cohen_h_h columns.
According to the documentation, these columns represent the lower and upper bounds of the 95% confidence interval for Cohen's h. For hypermethylated DMRs (cohen_h > 0), the reported cohen_h value generally lies between cohen_h_l and cohen_h_h, as expected.
However, for many hypomethylated DMRs (cohen_h < 0), I observe results such as:
cohen_h = -0.51135
cohen_h_l = 0.433795
cohen_h_h = 0.588909
In this case, the estimated Cohen's h does not appear to lie between the reported lower and upper confidence bounds. I also observe some hypomethylated DMRs where cohen_h_l is negative, while cohen_h_h remains positive.
cohen_h = -0.51135
cohen_h_l = - 0.433795
cohen_h_h = 0.588909
I have used this command
modkit dmr pair -a controlt5mC_cov15.bed.gz -b infected5mC_cov15.bed.gz -o 5mCgenes.dmr --regions-bed genes.bed --ref genomic.fna --base C --header
Could you please clarify:
How should cohen_h_l and cohen_h_h be interpreted for negative Cohen's h values?
Are these values reported as signed confidence interval bounds or as effect-size magnitudes?
What is the recommended way to filter hypomethylated DMRs using the confidence interval columns?
Thank you for your help.
Hello,
I am using modkit dmr and would like clarification on the interpretation of the cohen_h_l and cohen_h_h columns.
According to the documentation, these columns represent the lower and upper bounds of the 95% confidence interval for Cohen's h. For hypermethylated DMRs (cohen_h > 0), the reported cohen_h value generally lies between cohen_h_l and cohen_h_h, as expected.
However, for many hypomethylated DMRs (cohen_h < 0), I observe results such as:
cohen_h = -0.51135
cohen_h_l = 0.433795
cohen_h_h = 0.588909
In this case, the estimated Cohen's h does not appear to lie between the reported lower and upper confidence bounds. I also observe some hypomethylated DMRs where cohen_h_l is negative, while cohen_h_h remains positive.
cohen_h = -0.51135
cohen_h_l = - 0.433795
cohen_h_h = 0.588909
I have used this command
modkit dmr pair -a controlt5mC_cov15.bed.gz -b infected5mC_cov15.bed.gz -o 5mCgenes.dmr --regions-bed genes.bed --ref genomic.fna --base C --header
Could you please clarify:
How should cohen_h_l and cohen_h_h be interpreted for negative Cohen's h values?
Are these values reported as signed confidence interval bounds or as effect-size magnitudes?
What is the recommended way to filter hypomethylated DMRs using the confidence interval columns?
Thank you for your help.