marneylc / hrms-pub Goto Github PK

This is the README file for the hrms project that deals with the analysis of high resolution mass spectrometry data.

This code requires a list of name:mz pairs in a csv file. It is currently named LipidList.csv on my local machine. An example file should have been distributed with this code. If you would like a few example mass spectral data files go to https://drive.google.com/folderview?id=0B06AQDbcyIg8Q2ZFYm1RNmhnMUE&usp=sharing

You must use proteowizard to convert all raw spectra files to mzXML files. The filname variables refered to in the documentation are these mzXML files. The code attempts to work with low intensity peaks so make sure that proteowizard is not performing any thresholding or peak peaking prior to analysis.

Usage in R:

To perform analysis for one sample in R, call the following in the console:

setwd("path")
source("hrms.R")
main(filename,rtwin=c(20,70),mzwin=c(200,1000))

With "path" being the directory with files hrms.R, LipidList.csv, and mzXML files and where rtwin is the retention time window in sec and mzwin is the mzwindow in m/z units

If you want to run this for all data files in a directory run the following command:

files = list.files(".", pattern=".mzXML")
system.time(
for (i in 1:length(files)) {
  main(files[i],rtwin=c(20,70),mzwin=c(200,1000))
}
)
results <- signals_deviations() 

The last line will produce the csv files signals and deviations in the active directory, which puts multiple file's data together into two csv files plus returns the data.frames in the list variable "results".

If you are attempting to analyse lots of data with this, please checkout lcms.py in marneylc/LCMS_highthroughput for multithreading of both the conversion process and the hrms peak picking.

The separate file signals_deviations.R is made available so that multiple folders/plates of samples can be collated into a singal "signals.csv" file (and deviations.csv).

In the output, a value of "NA" means that no data exists in the mass spectrum for that target peak. A signal value of "0" means that there is mass spectral data collected for the target mass, but that it has a signal value of zero.

Note: somtimes when edited in windows, hrms.R can end up having invisible CR characters which causes errors in linux.

If you get a "No such file or directory" error run the following code in python

with open('hrms.R', 'rb+') as f:
    content = f.read()
    f.seek(0)
    f.write(content.replace(b'\r', b''))
    f.truncate()

For Usage in Terminal and Multithreading with python:

For use in multithreading, it is important that the R program Rscript.exe be located in the system path to be called from within python.

Currently, file conversion to mzXML or mzML is limited to windows environments, advice on this or any code would be greatly appreciated. A python script for this can be found at https://github.com/marneylc/LCMS_highthroughput/blob/master/python/convert_multithread.py . If you already have a clone of LCMS_highthroughput you can call it from within python if the lcms.py file is in your path.

from lcms import *

maxnumthreads = 8 # to use 8 cores
rawfiles = pygrep('raw','.')
for f in rawfiles:
t = mzXML_conv(f)
    t.start()
    while threading.activeCount() > maxnumthreads:
        time.sleep(0)

To run HRMS.R with multithreading run the following from within python::

from lcms import *
mzxmlfiles = pygrep('mzXML','.')
maxnumthreads = 8 # to use 8 cores

for f in mzxmlfiles:
    t = R_hrms(f)
    t.start()
    while threading.activeCount() > maxnumthreads:
        time.sleep(0)