Closed tonywu1999 closed 1 month ago
If making a singular QQ plot for all proteins, one idea was to standardize residuals (by dividing residuals by sample variance of the linear model) and overlay to see if the extra insight is valuable.
New subtasks - apparently there is a function called modelBasedQCPlots that does this already
Context
The linear mixed effects models that MSstats uses assume constant variance among its residuals. However, in MS-based proteomics, the distribution of intensity values tends to be right-skewed, which can lead to residuals of a linear mixed effects model being right-skewed (violating the constant variance assumption). As a result, MSstats applies log transformation on intensities in order to transform those values from a right-skew distribution to a normal distribution, leading to a better fit model.
New Subtasks
all
would make a QQ plot for each protein IDAcceptance Criteria
QQ plots have been generated