In pmbrophy/mzDataTable: Import and convert raw mass spectrometry data formats

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>"
)
#load the library and data.table
library(mzDataTable)

Introduction

The mzDataTable package aims to provide methods get .mzML/.mzXML data into data.table format through the use of mzR. There are three conversion functions provided:

1. mzML2dataTable reads an .mzML/.mzXML file into memory as a data.table
2. mzML2csv converts an .mzML/.mzXML file to a .csv file by first reading chunks of the .mzML/.mzXML file into memory as data.tables and sequentially writing them to the .csv file.
3. mzML2diskFrame converts an .mzML/.mzXML file to a disk.frame directory by first reading chunks of the .mzML/.mzXML file into memory as data.tables and sequentially writing them as disk.frame chunks. This provides efficent on-disk manipulation of raw mass spectrometry data.

Import and Conversion Functions

The mzML2dataTable Function: get data into memory

We can use the mock .mzML data provided in the mzDataTable package as an example or get real data from msdata.

dataPath <- msdata::proteomics(full.names = TRUE)[3]

#Read the .mzML file into memory as a data.table
dt <- mzML2dataTable(path = dataPath)

data.table::is.data.table(dt)

head(dt)

The mzML2diskFrame Function: get serialized data onto hard disk for on-disk manipulations

dskF_path <- paste0(getwd(), "/", basename(dataPath),".df")

#Create disk frame
dskF <- mzML2diskFrame(path = dataPath, diskFramePath = dskF_path)

#Connect to disk frame
#dskF <- disk.frame::disk.frame(path = dskF_path)
disk.frame::setup_disk.frame()
options(future.globals.maxSize = Inf)

Data Manipulation Functions

Once data are imported into R, there are some functions provided for basic manipulations:

1. getTIC calculates the total ion chromatogram
2. getBPI calculates the base peak intensity chromatogram
3. getXIC extracts summed ion signal intensity over a specified mass range
4. getSumSpectrum_dt produces a summed (and optionally normalized) mass spectrum

getTIC example: plot total ion chromatogram from MS level == 1, 2, and 3 individually

tic_dt <- getTIC(mzObj = dt[msLevel == 1], normalize = FALSE)
tic_df <- getTIC(mzObj = dskF[msLevel == 1], normalize = FALSE)

identical(tic_dt, tic_df)

tic_dt <- getTIC(mzObj = dt[msLevel == 1], normalize = TRUE)
tic_df <- getTIC(mzObj = dskF[msLevel == 1], normalize = TRUE)

identical(tic_dt, tic_df)

tic_dt <- getTIC(mzObj = dt, normalize = TRUE)
tic_df <- getTIC(mzObj = dskF, normalize = TRUE)

identical(tic_dt, tic_df)

plot(x = tic_dt$retentionTime, y = tic_dt$intensity, type = "l")

getBPI example: get base peak intensity

bpi_dt <- getBPI(mzObj = dt[msLevel == 1], normalize = FALSE)
bpi_df <- getBPI(mzObj = dskF[msLevel == 1], normalize = FALSE)

identical(bpi_dt, bpi_df)

bpi_dt <- getBPI(mzObj = dt[msLevel == 1], normalize = TRUE)
bpi_df <- getBPI(mzObj = dskF[msLevel == 1], normalize = TRUE)

identical(bpi_dt, bpi_df)

remove(bpi_dt, bpi_df)
bpi_dt <- getBPI(mzObj = dt, normalize = TRUE)
bpi_df <- getBPI(mzObj = dskF, normalize = TRUE)

identical(bpi_dt, bpi_df)

plot(x = bpi_dt$retentionTime, y = bpi_dt$intensity, type = "l")

getXIC example: extract m/z 376.2756 +/- 25 ppm

xic_dt <- getXIC(mzObj = dt[msLevel == 1], mz = 376.2756, ppmTol = 25, iStart = 1)

xic_dt <- getXIC(mzObj = dt[msLevel == 1], mz = 376.2756, ppmTol = 25)
xic_df <- getXIC(mzObj = dskF[msLevel == 1], mz = 376.2756, ppmTol = 25)

identical(xic_dt, xic_df)

xic_dt <- getXIC(mzObj = dt, mz = 376.2756, ppmTol = 25)
xic_df <- getXIC(mzObj = dskF, mz = 376.2756, ppmTol = 25)

identical(xic_dt, xic_df)

xic_dt <- getXIC(mzObj = dt, mz = 376.2756, ppmTol = 25, iStart = 10, iStop = 12)
xic_df <- getXIC(mzObj = dskF, mz = 376.2756, ppmTol = 25, iStart = 10, iStop = 12)

identical(xic_dt, xic_df)

xic_dt <- getXIC(mzObj = dt, mz = 376.2756, ppmTol = 25, tStart = 2730, tStop = 2750)
xic_df <- getXIC(mzObj = dskF, mz = 376.2756, ppmTol = 25, tStart = 2730, tStop = 2750)

identical(xic_dt, xic_df)

plot(x = xic_dt$retentionTime, y = xic_dt$intensity, type = "l")

getSumSpectrum_dt example: sum MS1 spectra

spec_norm <- getSumSpectrum(mzObj = dt[msLevel == 1], normalize = TRUE, isCentroid = FALSE)
spec <- getSumSpectrum(mzObj = dt[msLevel == 1], normalize = FALSE, isCentroid = FALSE)

spec_dt <- getSumSpectrum(mzObj = dt, normalize = FALSE, isCentroid = TRUE, ppmTol = 100)
spec_dskF <- getSumSpectrum(mzObj = dskF, normalize = FALSE, isCentroid = TRUE, ppmTol = 100)
identical(spec_dt, spec_dskF)

spec_dt <- getSumSpectrum(mzObj = dt, normalize = FALSE, isCentroid = TRUE, ppmTol = 100, iStart = 10, iStop = 20)
spec_dskF <- getSumSpectrum(mzObj = dskF, normalize = FALSE, isCentroid = TRUE, ppmTol = 100,  iStart = 10, iStop = 20)


spec <- getSumSpectrum(mzObj = dt, normalize = FALSE, isCentroid = TRUE, ppmTol = 100, tStart = 10, tStop = 20)
spec <- getSumSpectrum(mzObj = dskF, normalize = FALSE, isCentroid = TRUE, ppmTol = 100,  tStart = 10, tStop = 20)

plot(x = spec_dskF$mzGrid, y = spec_dskF$intensity, type = "h")

disk.frame::delete(df = dskF)

UNDERSTAND disk.frame OPERATION ORDER

#What happens? 1) read mz, seqNum, intensity from disk as data.table. 2) apply %between%
dskF_test <- dskF[seqNum %between% c(1,2), keep = c("mz", "seqNum", "intensity")]

#Is it the same as this?
dskF_test <- dskF[keep = c("mz", "seqNum", "intensity")]

dskF_test <- dskF_test[seqNum %between% c(1,2)]

#What happens with 1 vs 6 workers? 
#1) each worker reads in mz, seqNum, intensity
#2) each worker then subsets by %between%?

#So, in the case of a single worker... read in columns, subset, return subset to Global Env, and repeat?

pmbrophy/mzDataTable documentation built on June 6, 2020, 7:43 a.m.