EAC

eac.dmr <- readr::read_tsv("~/paper_elmer/ESCA_analysis_hg38/WGBS/ALL_HYPO_EAC_SIGNIFICANT_DMR_7.1.18",
                           col_names = F,col_types = "ciiicidddddiii")
colnames(eac.dmr) <- c("seqnames","start","end",
                       "length","strand","L",
                       "area", "beta","stat",
                       "pval","qval","index_start",
                       "index_end","L2")
eac.dmr$ID <- paste0(eac.dmr$seqnames,":",
                     eac.dmr$start,"-",
                     eac.dmr$end)

eac.dmr.gr <- makeGRangesFromDataFrame(eac.dmr,keep.extra.columns = T)

eac.probes <- readr::read_csv("~/paper_elmer/ESCA_analysis_hg38/escc_vs_esad_hg38/hyper/getMethdiff.hyper.probes.significant.csv",
                              col_types = readr::cols())
eac.pairs <- readr::read_csv("~/paper_elmer/ESCA_analysis_hg38/escc_vs_esad_hg38/hyper/getPair.hyper.pairs.significant.csv",
                             col_types = readr::cols())
colnames(eac.probes)[1] <- "Probe"
eac.probes <- left_join(eac.probes,probes.metadata)

## Joining, by = "Probe"

eac.probes.gr <- makeGRangesFromDataFrame(eac.probes, keep.extra.columns = T)

t(tibble(nb.sig.probes = length(eac.probes.gr),
         nb.sig.dmr = length(eac.dmr.gr),
         nb.overlapped.450K.probes = length(unique(queryHits(findOverlaps(eac.dmr.gr,probes.metadata.gr)))),
         nb.overlapped.450K.distal.probes = length(unique(queryHits(findOverlaps(eac.dmr.gr,distal.probes)))),
         nb.overlapped.sig.probes = length(unique(subjectHits(findOverlaps(eac.dmr.gr,eac.probes.gr)))),
         nb.overlapped.dmr = length(unique(queryHits(findOverlaps(eac.dmr.gr,eac.probes.gr)))),
         nb.overlapped.sig.probes.extended = length(unique(subjectHits(findOverlaps(eac.dmr.gr,
                                                                                    resize(eac.probes.gr,width = 501,fix = "center"))))),
         nb.overlapped.sig.dmr.extended = length(unique(queryHits(findOverlaps(eac.dmr.gr,
                                                                               resize(eac.probes.gr,width = 501,fix = "center")))))
))

##                                   [,1]
## nb.sig.probes                     1066
## nb.sig.dmr                        1446
## nb.overlapped.450K.probes          376
## nb.overlapped.450K.distal.probes   248
## nb.overlapped.sig.probes            89
## nb.overlapped.dmr                   70
## nb.overlapped.sig.probes.extended   95
## nb.overlapped.sig.dmr.extended      73

ESCC

escc.dmr <- readr::read_tsv("~/paper_elmer/ESCA_analysis_hg38/WGBS/HYPO_ESCC_SIGNIFICANT_DMR_7.1.18",
                            col_names = F,
                            col_types = "ciiicidddddiii")

colnames(escc.dmr) <- c("seqnames","start","end",
                        "length","strand","L",
                        "area", "beta","stat",
                        "pval","qval","index_start",
                        "index_end","L2")
escc.dmr$ID <- paste0(escc.dmr$seqnames,":",
                      escc.dmr$start,"-",
                      escc.dmr$end)
escc.dmr.gr <- makeGRangesFromDataFrame(escc.dmr,keep.extra.columns = T)

escc.probes <- readr::read_csv("~/paper_elmer/ESCA_analysis_hg38/escc_vs_esad_hg38/hypo/getMethdiff.hypo.probes.significant.csv",
                               col_types = readr::cols())
escc.pairs <- readr::read_csv("~/paper_elmer/ESCA_analysis_hg38/escc_vs_esad_hg38/hypo/getPair.hypo.pairs.significant.csv",
                              col_types = readr::cols())
colnames(escc.probes)[1] <- "Probe"
escc.probes <- left_join(escc.probes,probes.metadata)

## Joining, by = "Probe"

escc.probes.gr <- makeGRangesFromDataFrame(escc.probes, keep.extra.columns = T)

t(tibble(nb.sig.probes = length(escc.probes.gr),
         nb.sig.dmr = length(escc.dmr.gr),
         nb.overlapped.450K.probes = length(unique(queryHits(findOverlaps(escc.dmr.gr,probes.metadata.gr)))),
         nb.overlapped.450K.distal.probes = length(unique(queryHits(findOverlaps(escc.dmr.gr,distal.probes)))),
         nb.overlapped.sig.probes = length(unique(subjectHits(findOverlaps(escc.dmr.gr,escc.probes.gr)))),
         nb.overlapped.dmr = length(unique(queryHits(findOverlaps(escc.dmr.gr,escc.probes.gr)))),
         nb.overlapped.sig.probes.extended = length(unique(subjectHits(findOverlaps(escc.dmr.gr,
                                                                                    resize(escc.probes.gr,width = 501,fix = "center"))))),
         nb.overlapped.sig.dmr.extended = length(unique(queryHits(findOverlaps(escc.dmr.gr,
                                                                               resize(escc.probes.gr,width = 501,fix = "center")))))
))

##                                   [,1]
## nb.sig.probes                     1239
## nb.sig.dmr                         203
## nb.overlapped.450K.probes           63
## nb.overlapped.450K.distal.probes    39
## nb.overlapped.sig.probes            15
## nb.overlapped.dmr                   11
## nb.overlapped.sig.probes.extended   18
## nb.overlapped.sig.dmr.extended      12

DMR vs 450k

EAC

ESCC