---
title: "07 - Differentially methylated regions (1kb tiles)"
output: html_document
---

```{r setup, include=FALSE}
knitr::opts_chunk$set(eval = FALSE, message = FALSE, warning = FALSE)
library(methylKit); library(data.table)
```

Region-level, memory-efficient complement to the per-CpG DMLs. Tile the genome
into non-overlapping 1 kb windows (>= 3 covered CpGs each) and re-run the same
three contrasts.

```{r tile}
filt <- readRDS("methylRaw_filtered.rds")
tiles <- tileMethylCounts(filt, win.size = 1000, step.size = 1000, cov.bases = 3)
saveRDS(tiles, "tiles.rds")   # ~441,760 windows/sample
```

```{r dmr}
meta <- fread("sample_metadata.csv")
run_dmr <- function(tiles, meta, idx, treat, label, qv = 0.01, md = 15) {
  sub <- reorganize(tiles, sample.ids = as.list(meta$sampleID[idx]), treatment = treat)
  m   <- unite(sub, destrand = FALSE)
  dm  <- calculateDiffMeth(m, mc.cores = 4)
  all <- as.data.table(getData(dm))
  dmr <- all[qvalue < qv & abs(meth.diff) >= md]
  fwrite(all, paste0("dmr_", label, "_all.csv.gz")); fwrite(dmr, paste0("dmr_", label, ".csv"))
  data.table(contrast = label, windows = nrow(all), n_DMR = nrow(dmr),
             hyper = sum(dmr$meth.diff > 0), hypo = sum(dmr$meth.diff < 0))
}

d1 <- run_dmr(tiles, meta, 1:12,  c(rep(0,6),rep(1,6)), "pH_within_2N")
d2 <- run_dmr(tiles, meta, 13:24, c(rep(0,6),rep(1,6)), "pH_within_3N")
# ploidy main
sub <- reorganize(tiles, sample.ids = as.list(meta$sampleID), treatment = c(rep(0,12),rep(1,12)))
m <- unite(sub, destrand = FALSE); dm <- calculateDiffMeth(m, mc.cores = 4)
all3 <- as.data.table(getData(dm)); dmr3 <- all3[qvalue < 0.01 & abs(meth.diff) >= 15]
fwrite(all3, "dmr_ploidy_main_all.csv.gz"); fwrite(dmr3, "dmr_ploidy_main.csv")
d3 <- data.table(contrast="ploidy_main", windows=nrow(all3), n_DMR=nrow(dmr3),
                 hyper=sum(dmr3$meth.diff>0), hypo=sum(dmr3$meth.diff<0))

dmr_summary <- rbind(d1, d2, d3); fwrite(dmr_summary, "dmr_summary.csv")
dmr_summary[]
```

Expected: pH within 2N ~1,954 DMRs (mostly hyper); pH within 3N ~2,760 (mostly
hypo); ploidy main ~198. Reproduces the ploidy-dependent pH direction flip at
the region level.