1导入数据
library(xcms)
library(RColorBrewer)
library(ggplot2)
library(ggrepel)
library(stringr)
2多组数据
myfiles <- list.files(pattern = "^neg")
myfiles
name<- str_sub(basename(myfiles),11,-6)
name
pd <- data.frame(sample_name = sub(name, pattern = ".mzML",
replacement = "", fixed = TRUE),
sample_group = c(rep("FA", 6), rep("HFD", 6)),
stringsAsFactors = FALSE)
pd
data_raw <- readMSData(files = myfiles, pdata = new("NAnnotatedDataFrame", pd),mode = "onDisk")
3 色谱峰检测
3.1
3.2 检测features
xchr <- findChromPeaks(data_raw, param = CentWaveParam(snthresh = 5))
此时dda_data多了msFeatureData属性，检测到的色谱峰信息储存在
dda_data@msFeatureData[["chromPeaks"]]
dda_data@msFeatureData[["chromPeaks"]]%>% head()
或者
mz = 117.0194
mzr = mz + c(-0.01,0.01)
chromPeaks(xchr,mz = mzr)
# mz mzmin mzmax rt rtmin rtmax into intb maxo sn sample
#CP00387 117.0196 117.0166 117.0202 225.976 201.758 251.732 201143.331 199248.458 11735.4843 153 1
#CP00946 117.0190 117.0185 117.0202 651.321 631.142 684.479 328866.455 321557.847 16002.3513 44 1
#CP01897 117.0198 117.0179 117.0210 235.163 194.475 267.051 20846.146 17670.137 760.9054 7 2
data <- chromPeaks(xchr) |> data.frame()
i = 946#琥珀酸
mz = c(data$mzmin[i],data$mzmax[i])
rt = c(data$rtmin[i],data$rtmax[i])
mzr <- mz + c(-0.01, 0.01)
rtr <- rt + c(-20, 20)
提取两组琥珀酸色谱图
chr_raw <- chromatogram(data_raw, mz = mzr, rt = rtr)
group_colors <- paste0(brewer.pal(3, "Set1")[1:2], "60")
names(group_colors) <- c("FA", "HFD")
plot(chr_raw, col = group_colors[chr_raw$sample_group])
data_raw |>
filterRt(rtr)|>
filterMz(mzr)|>
plot(type = "XIC")
xchr <- findChromPeaks(chr_raw, param = CentWaveParam(snthresh = 2))
sample_colors <- group_colors[xchr$sample_group]
bg <- sample_colors[chromPeaks(xchr)[, "column"]]
plot(xchr, col = sample_colors, peakBg = bg)
设置参数
cwp <- CentWaveParam(snthresh = 5)
xdata <- findChromPeaks(data_raw, param = cwp)
设置参数
使用centWave算法对centroid模式的高分辨LC-MS进行色谱峰检测。centWave算法最适用于高分辨率 centroid模式 的LC/{TOF、OrbiTrap、FTICR}-MS数据。在第一阶段，该方法确定了感兴趣区域(ROI)，这些区域代表了LC/MS连续扫描时小于ppm m/z偏差的质量轨迹。 详细地说，从单个m/z开始，如果在下一次扫描(频谱)中发现的m/z，其与平均m/z的差异小于用户定义的m/z的ppm，则合并为一个ROI。 考虑到新加入的m/z值，ROI的平均m/z值也随之更新。
合并分裂峰
mpp <- MergeNeighboringPeaksParam(expandRt = 4)
xdata_pp <- refineChromPeaks(xdata, mpp)
参数
expandRt
numeric(1) defining by how many seconds the retention time window is expanded on both sides to check for overlapping peaks.
expandMz
numeric(1) constant value by which the m/z range of each chromatographic peak is expanded (on both sides!) to check for overlapping peaks.
ppm
numeric(1) defining a m/z relative value (in parts per million) by which the m/z range of each chromatographic peak is expanded to check for overlapping peaks.
minProp
numeric(1) between 0 and 1 representing the proporion of intensity to be required for peaks to be joined. See description for more details. The default (minProp = 0.75) means that peaks are only joined if the signal half way between then is larger 75% of the smallest of the two peak's "maxo" (maximal intensity at peak apex).
object
XCMSnExp object with identified chromatographic peaks.
param
MergeNeighboringPeaksParam object defining the settings for the method.
msLevel
integer defining for which MS level(s) the chromatographic peaks should be merged.
BPPARAM
parameter object to set up parallel processing. Uses the default parallel processing setup returned by bpparam(). See bpparam() for details and examples.
chr_ex <- chromatogram(xdata_pp, mz = mzr, rt = rtr)
chromPeaks(chr_ex)
sample_colors <- group_colors[chr_ex$sample_group]
plot(chr_ex, col = group_colors[chr_raw$sample_group], lwd = 2,
peakBg = sample_colors[chromPeaks(chr_ex)[, "sample"]])
plot(chr_ex, col = sample_colors, peakType = "rectangle",
peakCol = sample_colors[chromPeaks(chr_ex)[, "sample"]],
peakBg = NA)
## Extract a list of per-sample peak intensities (in log2 scale)
ints <- split(log2(chromPeaks(xdata_pp)[, "into"]),
f = chromPeaks(xdata_pp)[, "sample"])
boxplot(ints, varwidth = TRUE, col = sample_colors,
ylab = expression(log[2]~intensity), main = "Peak intensities")
grid(nx = NA, ny = NULL)
参考资料：
https://bioconductor.org/packages/release/bioc/html/xcms.html
LCMS data preprocessing and analysis with xcms (bioconductor.org)
