In ChemoSens/PTRMSR: Analysis of PTRMS data

\tableofcontents

This vignette shows how to use PTRMSR package to read .txt files and calculate some statistics.

Preliminary work

Installing PTRMSR packages

The package can be uploaded with the following instructions. It requires several external packages that should also beinstalled.

# remove.packages("PTRMSR)
# install.packages("devtools")
# install.packages("BiocManager")
# BiocManager::install("rhdf5") #<- update packages: no for faster
# BiocManager::install("MSnbase")#<- update packages: no for faster
library(rhdf5)
library(MSnbase)
library(devtools)
install_github("https://github.com/ChemoSens/PTRMSR")

General principle

In this package, the functions beginning by ptrv use a single dataset as input. The functions beginning by ptrvList use several files and a metaData file containing information about the different files. The objective of this paper is to use files produced by ptrViewer (or any other preprocessing of PTR-MS device returning files with ion, intensity and time as columns) then return curves and relevant statistics on the evaluations. Several biases can be corrected: (i) the breathing of the subject by calculating statistics by breathing cycle (use the parameter correction="cycle"),(ii) smoothing can be applied (instead of breathing cycle correction) (iii) blank period during the evaluation that can be removed from the intensities (use removeNoise = TRUE and noisePeriod=...). Some statistics are returned.

Loading the library

library(ggplot2)
library(gridExtra)
library(PTRMSR)

Pre-processing only one file

Loading a dataset

The first step is to load a dataset whose colnames are 'time', then the different ions. Each evaluation is a line of the dataset.

ptrv is an example object where the time column is 'RelTime', with two supplementary columns to remove (AbsTime and Cycle).

Correcting the breathing (or not)

ptrvIntensityByTime allows the breath to be corrected by calculating the breathing cycles than calculating a statistic per breathing cycle (funAggreagte='mean' for average or funAggregate="max" for maximum for example)

data(ptrv)
resptrv=ptrvIntensityByTime(dataset=ptrv,timeCol="RelTime",colToRemove=c("AbsTime","Cycle"),correction="cycle",referenceBreath="m69.06989..isoprene...Conc.",timePeriod=NULL,
             ions=NULL,funAggregate="mean",                     minExpi=NULL,maxInspi=NULL,smoothMethodBreath="MovingAverage",minimalDuration=2,forMinExpiDivideMaxIntBy=4,forMaxInspiDivideMaxIntBy=5,halfWindowSize=5, method="MAD",SNR=0)
names(resptrv)
resptrv$gg$p_cyclelimits
head(resptrv$res)

The obtained results contains a dataframe in a long format (as used in ggplot) whose colnames are time, intensity, ion and duration. Running the same function with the parameter correction="none" returns the raw data in a similar format without any correction of the breath.

resptrv_none=ptrvIntensityByTime(dataset=ptrv,timeCol="RelTime",colToRemove=c("AbsTime","Cycle"),correction="none")
head(resptrv_none$res)

Smoothing

The resulting data can be smoothed using ptrvSmooth function. It allows different methods of smoothing to be tested.

smoothed_df=ptrvSmooth(resptrv$res,method="Loess",spar=0.2,sameTime=FALSE)

ggplot(smoothed_df,aes(x=time,y=intensity,col=ion))+geom_line()+theme(legend.position="none")

Correcting by the noise

The time period can be corrected using ptrvRemoveNoise function

df_noiseCorrected=ptrvRemoveNoise(smoothed_df,timeBlank=c(0,30))

p=ggplot(df_noiseCorrected,aes(x=time,y=intensity,col=ion))+geom_line()+theme(legend.position="none")
#ggplotly(p)

Visualization of all corrections for one given ion

This part allows the different steps of pretreatemnt to be vizualized.

p1=ggplot(ptrv,aes(y=m45.03349...C2H4O.H....Conc.,x=RelTime))+geom_line()+ggtitle("Raw data")

resion=resptrv$res[resptrv$res[,"ion"]=="m45.03349...C2H4O.H....Conc.",]
resion_none=resptrv_none$res[resptrv_none$res[,"ion"]=="m45.03349...C2H4O.H....Conc.",]
p2=ggplot(resion,aes(x=time,y=intensity))+geom_line()+ggtitle("breath correction")


p3=ggplot(smoothed_df[smoothed_df[,"ion"]=="m45.03349...C2H4O.H....Conc.",],aes(x=time,y=intensity))+geom_line()+ggtitle("breath correction + smooth")
smoothed_raw_df=ptrvSmooth(resion_none,method="Loess",spar=0.2,sameTime=FALSE)
p4=ggplot(smoothed_raw_df,aes(x=time,y=intensity))+geom_line()+ggtitle("smoothed raw data")
p_noise=ggplot(df_noiseCorrected[df_noiseCorrected[,"ion"]=="m45.03349...C2H4O.H....Conc.",],aes(x=time,y=intensity))+geom_line()+ggtitle("noise correction")
grid.arrange(p1,p2,p3,p4,p_noise)

Managing several PTR-Viewer files with noise and breathing correction

Loading metadata and list of files to be processed

All the data files (.txt from ptrViewer) and the metadata file (.csv) should be in a single repository

repo="./../inst/extdata"
listFiles=list.files(repo,pattern="*.txt")
metaData=read.table(paste0(repo,"/metaData.csv"),sep=";",header=T)
head(metaData)

AUC calculations of several PTR-Viewer files with noise and breathing correction

This function allows statistics to be calculated for each file after breathing correction

setwd(repo)
res_auc=ptrvListIntensityByTime(listFiles=listFiles,
                                metaData=metaData,correction="cycle", ions=c("m31.0183", "m31.0259", "m33.0324"))
head(res_auc$listRes)

Parameters can be added using smoothing=TRUE then smoothing parameters and removeNoise=TRUE then removeNoise parameters (time) The results can be saved into csv files

write.table(file="auc.csv",sep=";",res_auc$listRes,row.names=F)

The stat option can be used to select the statistic.

repo="./../inst/extdata"
res_tmax=ptrvListIntensityByTime(listFiles=listFiles,metaData=metaData,ions=c("m31.0183", "m31.0259",                                                "m33.0324"),correction="cycle",stat="tmax")

The breathing cycles can be observed with this command.

grid.arrange(grobs=res_tmax$cycleLimits[1:4])

Documentations on the functions are available in the R documentation

?detectCycle
?ptrvIntensityByTime # for only one dataset, 
?ptrvIntensity # for summarizing the times of one dataset
?ptrvSmooth
?ptrvRemoveNoise
?ptrvListIntensityByTime # for only one dataset, 
?ptrvListIntensity # for summarizing several datasets

ChemoSens/PTRMSR documentation built on June 15, 2025, 10:40 a.m.