[BioC] IPPD model fit

[Yuri Galachyants]

Hi all,

I am trying to use IPPD package to process peptide spectra from
MALDI-TOF Brucker Ultraflextreme instrument. However, the model
function seems to be poorly fit the real spectrum. Basically, I passed
through the IPPD tutorial when analyzing data. The R code is presented
below.

library(IPPD)

pl <- read.table("25-up-01_0_E14_1.txt", header=TRUE)
x<-pl[,1]
y<-pl[,2]
y<-y[x>=500&x<=2500]
x<-x[x>=500&x<=2500]

gauss<-fitModelParameters(
x,y, model="Gaussian",
fitting=c("model"),
formula.sigma=formula(~mz),
control=list(window=6, threshold=200)
)

emg<-fitModelParameters(
x,y, model="EMG", fitting="model",
formula.alpha=formula(~mz),
formula.sigma=formula(~mz),
formula.mu=(~1),
control=list(window=6, threshold=200)
)

mse.emg<-data.frame(
mse=emg at peakfitresults[,2]/emg at peakfitresults[,1],
peakshape=rep('EMG', nrow(emg at peakfitresults))
)
mse.gauss<-data.frame(
mse=gauss at peakfitresults[,2]/gauss at peakfitresults[,1],
peakshape=rep('Gaussian', nrow(gauss at peakfitresults))
)

mses<-rbind(mse.gauss,mse.emg)

with(mses, boxplot(mse ~ peakshape, ylab='MSE'))
dev.copy(png, 'MSE.png')
dev.off()

visualize(gauss)
dev.copy(png, 'gauss_bestfit.png')
dev.off()

visualize(gauss, type='model', modelfit=TRUE)
dev.copy(png, 'gauss_regression.png')
dev.off()

visualize(emg)
dev.copy(png, 'emg_bestfit.png')
dev.off()

visualize(emg, type='model', modelfit=TRUE, parameters=c('alpha','sigma'))
dev.copy(png, 'emg_regression.png')
dev.off()

emglist<-getPeaklist(x, y, model='EMG', model.parameters=emg,
loss='L2', trace=FALSE, control.localnoise=list(factor.place=2),
control.basis=list(charges=c(1,2)),
control.postprocessing=list(ppm=200))
threshold(emglist, threshold=3, refit=TRUE, trace=FALSE)

visualize(emglist, x,y,lower=1089, upper=1096)
dev.copy(png, 'EMG_processed.png')
dev.off()


Model fitting statistics are in attachment.
Can anybody suggest the points to work with for refining the model?

-- 

Best regards, Yuri.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: EMG_processed.png
Type: image/png
Size: 24451 bytes
Desc: not available
URL: <https://stat.ethz.ch/pipermail/bioconductor/attachments/20140120/487cd244/attachment.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: MSE.png
Type: image/png
Size: 4678 bytes
Desc: not available
URL: <https://stat.ethz.ch/pipermail/bioconductor/attachments/20140120/487cd244/attachment-0001.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: gauss_regression.png
Type: image/png
Size: 7990 bytes
Desc: not available
URL: <https://stat.ethz.ch/pipermail/bioconductor/attachments/20140120/487cd244/attachment-0002.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: gauss_bestfit.png
Type: image/png
Size: 7140 bytes
Desc: not available
URL: <https://stat.ethz.ch/pipermail/bioconductor/attachments/20140120/487cd244/attachment-0003.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: emg_regression.png
Type: image/png
Size: 9775 bytes
Desc: not available
URL: <https://stat.ethz.ch/pipermail/bioconductor/attachments/20140120/487cd244/attachment-0004.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: emg_bestfit.png
Type: image/png
Size: 8262 bytes
Desc: not available
URL: <https://stat.ethz.ch/pipermail/bioconductor/attachments/20140120/487cd244/attachment-0005.png>