I am currently comparing methylation calls from deepsignal and nanopolish using DSS. Using the showOneDMR function, this plot is generated:

Note the axis labels for the read depth.

My R script:

library(DSS)

nano = read.table(file = 'nanopolish_methylation_freq.tsv', sep = '\t', header = TRUE)
deep = read.table(file = 'deepsignal_fast5s_single.freq-perCG-raw.tsv', sep = '\t', header = FALSE)

colnames(deep) <- c('chromosome','pos','strand','0_based_pos',
                    'prob_unmethy_sum','prob_methyl_sum',
                    'count_modified','count_unmodified',
                    'coverage','mod_freq','k_mer')

keeps <- c("chromosome", "start", "called_sites", "called_sites_methylated")
nano = nano[keeps]
keeps <- c("chromosome", "0_based_pos", "coverage", "count_modified")
deep = deep[keeps]

names(nano)[names(nano) == "start"] <- "pos"
names(nano)[names(nano) == "chromosome"] <- "chr"
names(nano)[names(nano) == "called_sites"] <- "N"
names(nano)[names(nano) == "called_sites_methylated"] <- "X"

names(deep)[names(deep) == "0_based_pos"] <- "pos"
names(deep)[names(deep) == "chromosome"] <- "chr"
names(deep)[names(deep) == "coverage"] <- "N"
names(deep)[names(deep) == "count_modified"] <- "X"


BSobj = makeBSseqData(list(nano, deep), 
                      c("nano", "deep"))[1:10000]

dmlTest = DMLtest(BSobj, group1 = c("nano"),group2 = c("deep"), smoothing = TRUE)

dmrs = callDMR(dmlTest, p.threshold = 0.01)

showOneDMR(dmrs[1,], BSobj)

Thanks for catching the bug. This is caused by the low sequencing depth. Apparently I didn't set the y-axis labels for read depth correctly. I have updated the package, but the changes will appear on bioc after a while. Can you send me your email, so that I can give you a new package to try? My email is hao.wu@emory.edu.

Regards, Hao Wu