LC Observations - Retention Time Stability
2024-10-31
Abstract
This document extends the work concerning the HPLC systems presented in (2016). We refer to the original manuscript for a description of the theory behind applications.
Package
msqc1 1.34.0
1 Introduction
This vignette file is thought to demonstrate the retention time stability of the msqc1_8rep and msqc1_dil data contained in the msqc1 R package.
1.1 Problem:
Retention Time (RT) form different LC-MS platforms are often not comparable (duration, offset).
1.2 Solution: Normalization
The iRT concept [proposed in pmid22577012 for the RT normalization is used. Instead of using the iRT-peptides MSQC1-peptides are used as reference peptides between the different chromatographic settings and systems.
The normalization is done by applying the following steps:
- define a reference platform (QTRAP)
- for each platform a linear model is build (in the following we use the
stats::lmfunction) which maps the RT space of the corresponding platform into reference platform RT space (by using thestats::predict.lmmethod) - scale the normalized RT values into [0, 1]
- visualize the normalized RT values
The code below shows the used R functions for the applied RT normalization.
msqc1:::.normalize_rt## function (S, S.training, reference_instrument = "Retention.Time.QTRQP")
## {
## S.normalized <- S
## for (instrument in unique(S.normalized$instrument)) {
## i <- paste("Retention.Time", instrument, sep = ".")
## rt.out <- S.training[, reference_instrument]
## rt.in <- S.training[, i]
## S.fit <- lm(rt.out ~ rt.in)
## S.normalized[S.normalized$instrument == instrument, "Retention.Time"] <- predict(S.fit,
## data.frame(rt.in = S.normalized[S.normalized$instrument ==
## instrument, "Retention.Time"]))
## }
## S.normalized.min <- min(S.normalized$Retention.Time, na.rm = TRUE)
## S.normalized.delta <- max(S.normalized$Retention.Time, na.rm = TRUE) -
## S.normalized.min
## S.normalized$Retention.Time <- (S.normalized$Retention.Time -
## S.normalized.min)/S.normalized.delta
## return(S.normalized)
## }
## <bytecode: 0x6545daefbf80>
## <environment: namespace:msqc1>The reshape_rt method is used for reshaping the data from long to wide format which ease the the model building in the msqc1:::.normalize_rt method above.
msqc1:::.reshape_rt## function (S, peptides = peptides, plot = TRUE, ...)
## {
## S <- S[grep("[by]", S$Fragment.Ion), ]
## S <- S[S$Peptide.Sequence %in% peptides$Peptide.Sequence,
## ]
## S <- aggregate(Retention.Time ~ Peptide.Sequence * instrument,
## FUN = mean, data = S)
## S <- droplevels(S)
## S.training <- reshape(S, direction = "wide", v.names = "Retention.Time",
## timevar = c("instrument"), idvar = "Peptide.Sequence")
## if (plot == TRUE) {
## pairs(S.training[, 2:6], pch = as.integer(S.training$Peptide.Sequence),
## col = as.integer(S.training$Peptide.Sequence), lower.panel = NULL,
## ...)
## }
## return(S.training)
## }
## <bytecode: 0x6545da98fa60>
## <environment: namespace:msqc1>The following R code listings displays some helper functions designed to ease the visualization of the msqc1_8rep and msqc1_dil RT values.
msqc1:::.plot_rt_8rep## function (S, peptides, ...)
## {
## .figure_setup()
## S <- S[grep("[by]", S$Fragment.Ion), ]
## S <- S[S$Peptide.Sequence %in% peptides$Peptide.Sequence,
## ]
## S <- aggregate(Retention.Time ~ Peptide.Sequence * File.Name.Id *
## instrument * relative.amount * Isotope.Label.Type, FUN = mean,
## data = S)
## S <- droplevels(S)
## xyplot(Retention.Time ~ File.Name.Id | Isotope.Label.Type *
## instrument, data = S, layout = c(10, 1), group = S$Peptide.Sequence,
## auto.key = list(space = "right", points = TRUE, lines = FALSE,
## cex = 1), ...)
## }
## <bytecode: 0x6545dac145a0>
## <environment: namespace:msqc1>msqc1:::.plot_rt_dil## function (S, peptides, ...)
## {
## .figure_setup()
## S <- S[grep("[by]", S$Fragment.Ion), ]
## S <- S[S$Peptide.Sequence %in% peptides$Peptide.Sequence,
## ]
## S <- aggregate(Retention.Time ~ Peptide.Sequence * File.Name *
## instrument * relative.amount * Isotope.Label.Type, FUN = mean,
## data = S)
## S <- droplevels(S)
## xyplot(Retention.Time ~ relative.amount | Isotope.Label.Type *
## instrument, data = S, layout = c(10, 1), group = S$Peptide.Sequence,
## scales = list(x = list(rot = 45, log = TRUE, at = sort(unique(S$relative.amount)))),
## auto.key = list(space = "right", points = TRUE, lines = FALSE,
## cex = 1), ...)
## }
## <bytecode: 0x6545dac88430>
## <environment: namespace:msqc1>2 RT plot
2.1 8 Technical Replicates RT plot
Prepare the training data for the linear model.
S.training.8rep <- msqc1:::.reshape_rt(msqc1_8rep, peptides=peptides, cex=2)
8 rep Scatterplot Matrices Plot - Color and icon type indicates the differnet peptides. On an idea plot all values would be on one line.
msqc1:::.plot_rt_8rep(msqc1_8rep, peptides=peptides, xlab='Replicate Id')
8 replicate retention time - The graph displays the raw retention time (in minutes) versus the replicate Id of each sample. Each panel displays one LC-MS platform. On some platforms the loading phase was recorded (TRIPLETOF, QTRAP) while on the other platforms not.
The following code will apply the retention time normalization for the msqc_8rep data:
msqc1_8rep.norm <- msqc1:::.normalize_rt(msqc1_8rep, S.training.8rep,
reference_instrument = 'Retention.Time.QTRAP')
msqc1:::.plot_rt_8rep(msqc1_8rep.norm,
peptides=peptides,
xlab='Replicate Id',
ylab='Normalized Retention Time')
Normalized 8 replicate retention time - The graphics displays the normalized retention time for each peptide (heavy and light) across all platforms. Apart from the TAENFR peptide all peptides show excellent elution time stability.
2.2 Dilution Series RT plot
S.training.dil <- msqc1:::.reshape_rt(msqc1_dil, peptides=peptides, cex=2)
Dilution Scatterplot Matrices Plot - Color and icon type indicates the differnet peptides. On an idea plot all values would be on one line.
The graph in displays the raw RT (in minutes) versus the relative amount (x-axis in log scale) of each sample. Each panel displays one LC-MS platform for each isotope label typ. The graph on the bottom displays the normalized RT for each peptide (heavy and light). Note: Since we have used different LC settings for the dilution data as we have used for the 8 rep data, we had to rebuild the linear models for the normalization.
msqc1:::.plot_rt_dil(msqc1_dil, peptides, xlab='Replicate Id', ylab='Raw Retention Time')
msqc1_dil.norm <- msqc1:::.normalize_rt(msqc1_dil,
S.training.dil,
reference_instrument = 'Retention.Time.QTRAP')
msqc1:::.plot_rt_dil(msqc1_dil.norm, peptides=peptides, ylab="Normalized Retention Time")
2.3 LC Gradient Comparison
The graphs compare the LC gradient of each platform by plotting the normalized RT values again the raw RT values for the msqc1_8rep (left) and msqc1_dil (right) data.
3 Session information
sessionInfo()## R version 4.4.1 (2024-06-14)
## Platform: x86_64-pc-linux-gnu
## Running under: Ubuntu 24.04.1 LTS
##
## Matrix products: default
## BLAS: /home/biocbuild/bbs-3.20-bioc/R/lib/libRblas.so
## LAPACK: /usr/lib/x86_64-linux-gnu/lapack/liblapack.so.3.12.0
##
## locale:
## [1] LC_CTYPE=en_US.UTF-8 LC_NUMERIC=C
## [3] LC_TIME=en_GB LC_COLLATE=C
## [5] LC_MONETARY=en_US.UTF-8 LC_MESSAGES=en_US.UTF-8
## [7] LC_PAPER=en_US.UTF-8 LC_NAME=C
## [9] LC_ADDRESS=C LC_TELEPHONE=C
## [11] LC_MEASUREMENT=en_US.UTF-8 LC_IDENTIFICATION=C
##
## time zone: America/New_York
## tzcode source: system (glibc)
##
## attached base packages:
## [1] stats graphics grDevices utils datasets methods base
##
## other attached packages:
## [1] msqc1_1.34.0 lattice_0.22-6 BiocStyle_2.34.0
##
## loaded via a namespace (and not attached):
## [1] cli_3.6.3 knitr_1.48 rlang_1.1.4
## [4] magick_2.8.5 xfun_0.48 highr_0.11
## [7] jsonlite_1.8.9 htmltools_0.5.8.1 tinytex_0.53
## [10] sass_0.4.9 rmarkdown_2.28 grid_4.4.1
## [13] evaluate_1.0.1 jquerylib_0.1.4 fastmap_1.2.0
## [16] yaml_2.3.10 lifecycle_1.0.4 bookdown_0.41
## [19] BiocManager_1.30.25 compiler_4.4.1 Rcpp_1.0.13
## [22] digest_0.6.37 R6_2.5.1 magrittr_2.0.3
## [25] bslib_0.8.0 tools_4.4.1 cachem_1.1.0References
Kockmann, Tobias, Christian Trachsel, Christian Panse, Åsa Wåhlander, Nathalie Selevsek, Jonas Grossmann, Witold E. Wolski, and Ralph Schlapbach. 2016. “Targeted proteomics coming of age - SRM, PRM and DIA performance evaluated from a core facility perspective.” PROTEOMICS. http://onlinelibrary.wiley.com/doi/10.1002/pmic.201500502/full.