Merge pull request #131 from yixuan-chen-elisa/read.modkit

Create a function read.modkit to read in methyl Bed files created by modkit to Bsseq object

DESCRIPTION

@@ -60,6 +60,7 @@ Collate:
     combine.R
     read.bismark.R
     read.modbam2bed.R
+    read.modkit.R
     BSmooth.R
     BSmooth.tstat.R
     dmrFinder.R

NAMESPACE

@@ -81,7 +81,7 @@ export("BSseq", "getMeth", "getCoverage", "getBSseq", "getStats",
        "combineList", "strandCollapse",
        "plotRegion", "plotManyRegions",
        "read.bismark",
-       "read.modbam2bed",
+       "read.modbam2bed", "read.modkit",
        "poissonGoodnessOfFit", "binomialGoodnessOfFit",
        "data.frame2GRanges", "BSseqTstat",
        "BSseqStat",

R/BSseq-class.R

@@ -165,8 +165,8 @@ BSseq <- function(M = NULL, Cov = NULL, Filtered = NULL, coef = NULL, se.coef =
 
     # Construct BSseq object ---------------------------------------------------
 
-    assays <- SimpleList(M = M, Cov = Cov, Filtered = Filtered, coef = coef,
-                         se.coef = se.coef)
+    assays <- SimpleList(M = M, Cov = Cov, Filtered = Filtered,
+                         coef = coef, se.coef = se.coef)
     assays <- assays[!S4Vectors:::sapply_isNULL(assays)]
     se <- SummarizedExperiment(
         assays = assays,

R/read.modkit.R

@@ -0,0 +1,78 @@
+read.modkit = function(files,
+                       colData = NULL,
+                       rmZeroCov = FALSE,
+                       strandCollapse = TRUE){
+  gr_list = list()
+  sampleNames = sub("\\.bed$", "", basename(files))
+  if (!is.null(colData)){
+    rownames(colData) <- sampleNames
+  }
+
+  for (i in seq_along(files)){
+    data = read.table(files[i], header = FALSE, sep="\t",
+                      stringsAsFactors=FALSE, quote="")
+
+    gr = GRanges(seqnames = data$V1,
+                 ranges = IRanges(start = data$V2+1, end = data$V3))
+    mcols(gr)$m = data$V12
+    mcols(gr)$cov = data$V12 + data$V13
+    mcols(gr)$filter = data$V16
+    if (any(data$V14 != 0)){
+        mcols(gr)$other_mod = data$V14
+        mcols(gr)$cov_other_mod = data$V14 + data$V13}
+    names(gr) = sampleNames[i]
+    gr_list[[sampleNames[i]]] = gr
+  }
+
+  overlap_gr = Reduce(subsetByOverlaps, gr_list)
+
+  m_cov_list = lapply(gr_list, function(gr){
+    overlap_data = gr[gr %over% overlap_gr]
+    if (!is.null(gr$other_mod)) {
+      data.frame(m = overlap_data$m, cov = overlap_data$cov,
+                 filter = overlap_data$filter,
+                 other_mod = overlap_data$other_mod,
+                 cov_other_mod = overlap_data$cov_other_mod)
+      } else {
+      data.frame(m = overlap_data$m, cov = overlap_data$cov,
+                 filter = overlap_data$filter)
+      }
+    })
+
+  m = do.call(cbind, lapply(m_cov_list, `[[`, "m"))
+  cov = do.call(cbind, lapply(m_cov_list, `[[`, "cov"))
+  filter = do.call(cbind, lapply(m_cov_list, `[[`, "filter"))
+  other_mod = do.call(cbind, lapply(m_cov_list, `[[`, "other_mod"))
+  cov_other_mod = do.call(cbind, lapply(m_cov_list, `[[`, "cov_other_mod"))
+
+  bsseq_objs = list()
+  bsseq_obj_m = BSseq(M = as.matrix(m), Cov = as.matrix(cov),
+                    Filtered = as.matrix(filter),
+                    coef = NULL, se.coef = NULL,
+                    pos = start(overlap_gr), trans = NULL,
+                    parameters = NULL, pData = colData, gr = NULL,
+                    chr = as.vector(seqnames(overlap_gr)),
+                    sampleNames = sampleNames, rmZeroCov = rmZeroCov)
+
+  bsseq_objs[[1]] = bsseq_obj_m
+
+  if (!is.null(other_mod)){
+      bsseq_obj_other_mod = BSseq(M = as.matrix(other_mod),
+                                  Cov = as.matrix(cov_other_mod),
+                                  Filtered = as.matrix(filter),
+                                  coef = NULL, se.coef = NULL,
+                                  pos = start(overlap_gr), trans = NULL,
+                                  parameters = NULL, pData = colData, gr = NULL,
+                                  chr = as.vector(seqnames(overlap_gr)),
+                                  sampleNames = sampleNames, rmZeroCov = rmZeroCov)
+
+      bsseq_objs[[2]] = bsseq_obj_other_mod
+    }
+
+  if (strandCollapse){
+      for (i in 1:length(bsseq_objs)) {
+      bsseq:::strandCollapse(bsseq_objs[[i]])}
+  }
+
+  return(bsseq_objs)
+}

man/read.modbam2bed.Rd

@@ -1,5 +1,3 @@
-% Generated by roxygen2: do not edit by hand
-% Please edit documentation in R/read.modbam2bed.R
 \name{read.modbam2bed}
 \alias{read.modbam2bed}
 \title{Construct BSseq objects from nanopore BED files}

man/read.modkit.Rd

@@ -0,0 +1,46 @@
+\name{read.modkit}
+\alias{read.modkit}
+\title{Construct BSseq objects from nanopore BED files}
+\description{
+Construct BSseq objects from nanopore BED files
+}
+\usage{
+read.modkit(
+  files,
+  colData = NULL,
+  rmZeroCov = FALSE,
+  strandCollapse = TRUE
+)
+}
+\arguments{
+  \item{files}{vector, BED files}
+  \item{colData}{data frame, phenotypic data with samples as rows and variables as columns}
+  \item{rmZeroCov}{A logical (1) indicating whether methylation loci that have zero coverage in all samples be removed}
+  \item{strandCollapse}{A logical (1) indicating whether stand-symmetric methylation loci (i.e. CpGs) should be collapsed across strands}
+}
+
+\value{
+BSseq objects
+}
+\details{
+This function reads in nanopore sequencing modified BED files
+to Bsseq objects. Nanopore sequencing data (i.e. aggregated modified base
+counts) is stored in modified-base BAM files. These modified-base BAM files
+are converted to bedMethyl (BED) files using \href{https://github.com/nanoporetech/modkit}{modkit}.
+
+\subsection{Details for modkit}{
+  Modkit outputs modified reads, unmodified reads, ambiguous modification reads (reads where the probability was below the threshold and usually failing the lowest 10th percentile), and other modified reads.}
+
+\subsection{modkit to Bsseq object}{
+  After creating BED files using modkit, the BED files are read in and the Bsseq object is constructed via \code{read.modkit()} function. The function reads in BED files, extract genomic regions, methylation, coverage, ambiguous modification status data and sample information and then construct Bsseq object using \code{BSseq} function within the package. If there are other modified reads, two Bsseq objects are constructed with the first one using major modified reads and their corresponding coverage and the second one using the other modified reads and their corresponding coverage.}
+}
+
+\examples{
+# only one modification and one BSseq object is constructed.
+files <- c(system.file("extdata/modkit/chr22.HG002.top1000.bed.gz", package = "bsseq"))
+bsseq_nano <- bsseq::read.modkit(files, rmZeroCov = FALSE, strandCollapse=FALSE)
+
+# there is other modification, so two BSseq objects are constructed.
+files <- c(system.file("extdata/modkit/chr22.HG002.top1000.other_mod.bed.gz",package = "bsseq"))
+bsseq_nano <- bsseq::read.modkit(files, rmZeroCov = FALSE, strandCollapse=FALSE)
+}

tests/testthat/test_read.modkit.R

@@ -0,0 +1,13 @@
+context("read.modkit")
+
+# TODO: Re-factor read.modkit() and update tests accordingly
+test_that("read.modkit() works for BED files", {
+    infile <- system.file("extdata", "modkit/chr22.HG002.top1000.other_mod.bed.gz",
+                          package = "bsseq")
+    bsseq <- read.modkit(files = infile,
+                         colData = NULL,
+                         rmZeroCov = FALSE,
+                         strandCollapse = TRUE)
+
+    lapply(bsseq, function(x) {expect_is(x, "BSseq")})
+})