Compare the constructed spectrum with experimental native MS spectrum
Here we compare the constructed result with the experimental native MS spectrum.
Include the charge state envelope
To avoid artifacts possibly induced by spectra deconvolution, the constructed intact protein spectra are further processed to include a charge state envelope. In this regard, it can be directly compared with the unprocessed native MS data.
The relative abundance of each charge state is estimated based on native MS spectrum. The raw native MS spectrum is exported directly from Xcalibur and saved as txt format.)
toPlot <- generatePseudoGaussianSpectrum(
x=as.numeric(names(simulated))+backboneMass,
y=100/max(simulated)*simulated,
sd=5,
xlim=c(40000, 50000))
colours <- c("Simulated"="seagreen", "NativeMS"="darkgoldenrod3")
plot(x=(toPlot[[1]]+12)/12, y=toPlot[[2]]*0.3, type="l", lwd=1,xlab="m/z",
ylab="abundance", col=colours[1], yaxs="i",
xlim=c(2900,4300), ylim=c(-100,100))
lines(x=(toPlot[[1]]+13)/13, y=toPlot[[2]], col=colours[1])
lines(x=(toPlot[[1]]+14)/14, y=toPlot[[2]]*0.75, col=colours[1])
lines(x=(toPlot[[1]]+15)/15, y=toPlot[[2]]*0.1, col=colours[1])
fetuinFn <- file.path(system.file("extdata", package="glycoNativeMS"),
"Fetuin_nativeMS.txt")
data2 <- read.table(fetuinFn, header=FALSE, sep="\t")
data2 <- transform(data2 , V2=-V2/max(V2)*100)
lines(x=data2[[1]],
y=data2[[2]],
type="h", lwd=1,
col=colours[2])
legend("topright", legend=names(colours), lty=1, col=colours)
Calculate the correlation score
A Pearson correlation is calculated to assess the similarity of simulated and experimental spectra.
#Pearson crrelation
experimentalSpectrum <- data.frame(mass=data2[[1]],
abundance=-data2[[2]])
foo <- as.numeric(names(simulated))+backboneMass
fooAbundance <- 100/max(simulated)*simulated
simulatedSpectrum <- data.frame(mass=c((foo+15)/15,(foo+12)/12,
(foo+13)/13, (foo+14)/14),
abundance=c(fooAbundance*0.1, fooAbundance*0.3,
fooAbundance, fooAbundance*0.75))To further reduce the instrument noise level, raw native MS spectra are pre-processed by binning adjacent data points into a defined bin width, the constructed spectrum is processed using the same defined range for comparison. Subsequently, the relative abundances within a bin are summed up.
For Fetuin, m/z range 3000-4500 is considered. The bin width is decided by the instrument resolution and the noise level of the native MS spectrum (in this case a bin width of 4 is used)
ranges <- seq(3000, 4500, by=4)
experimentalAggregated <- tapply(experimentalSpectrum$abundance,
cut(experimentalSpectrum$mass, ranges),
sum, na.rm=TRUE)
simulatedAggregated <- tapply(simulatedSpectrum$abundance,
cut(simulatedSpectrum$mass, ranges),
sum, na.rm=TRUE)
## pearson correlation
cor.test(experimentalAggregated, simulatedAggregated, use="complete.obs")
#>
#> Pearson's product-moment correlation
#>
#> data: experimentalAggregated and simulatedAggregated
#> t = 30.999, df = 282, p-value < 2.2e-16
#> alternative hypothesis: true correlation is not equal to 0
#> 95 percent confidence interval:
#> 0.8498565 0.9032289
#> sample estimates:
#> cor
#> 0.8792737