Skip to main content
Waters

What is the contribution rate of each LCMS peak to the loading plot in multivariate analysis in EZinfo - WKB53889

Article number: 53889

ENVIRONMENT

  • EZinfo
  • Progenesis QI
  • Progenesis QI for Proteomics
  • UNIFI
  • MarkerLynx

ANSWER

The individual LC or MS data peaks contribute only indirectly to the scores plot. The scores plot shows the variables that are most different between the various sample groups. If you hover over the “Scores” button in the Plots and Spreadsheets tab you’ll see this explanation.

“Scores t[1] and t[2] are the most important indices in summarizing the dataset. The plot gives a picture of the data. The points in the plot are the observations in the data. Observations near each other in the plot are probably similar and observations far away from each other are dissimilar. The plot shows the presence of atypical observations, groups similarities, trends, and other patterns in the data.”


In the scores plot “Show model statistics” t[1] (the x axis) and t[2] (the y axis) are the lines of best fits drawn though a multidimensional dimensional scatter plot of the differences between the data and the mean of the data. Those lines represent the maximum differences between the data in different groups within the sample set. R2X[1] and R2X[2] are the R squared value for those lines, i.e. a statistical measure of how well those lines fit the multidimensional plot of the data (it would be clearer if this was written R2X[1]).

This means that if there are peaks at a particular retention time that are much more intense in samples from the "treatment 1" group than in samples from the "control" group, then those peaks will contribute much more in terms of differentiating between treatment 1 and control.
The Hotelling’s T2 value represents the 95% confidence limit of the scores plot. Anything outside that ellipse is a strong outlier.

The goodness of fit plot is a graphical representation of how well the model of the data represented by the Scores plot actually fits the data.

For a full explanation, see https://umetrics.com/sites/default/f..._parti-3_0.pdf.

ADDITIONAL INFORMATION

NOTE : MarkerLynx is discontinued, and Waters no longer sells either MarkerLynx or the associated EZinfo 2.0. However, the above answer is equally applicable to EZinfo 2 or 3.

id53889,

Not able to find a solution? Click here to request help.