Table II. Partial areas under the curve (pAUC) for a false positive rate (FPR) < 0.1 for three standard approaches and our robust ridge method by comparing conditions 6B-6A, 6C-6A and 6C-6B in the CPTAC spike-in study. MaxLFQ+Perseus: protein-level analysis consisting of MaxLFQ normalization followed by t-tests in Perseus, MaxLFQ+limma: protein-level analysis consisting of MaxLFQ normalization followed by limma analysis, LM: peptide-based linear regression model containing treatment, peptide and instrument effects, RR: peptide-based ridge regression model containing treatment, peptide and instrument effects with empirical Bayes variance estimator and M-estimation with Huber weights.

Comparison	MaxLFQ+Perseus	MaxLFQ+limma	LM	RR
6B-6A	0.061	0.075	0.083	0.091
6C-6A	0.076	0.088	0.096	0.097
6D-6A	0.077	0.089	0.098	0.097
6E-6A	0.076	0.089	0.097	0.097
6C-6B	0.081	0.086	0.096	0.097
6D-6B	0.082	0.086	0.096	0.096
6E-6B	0.081	0.087	0.096	0.096
6D-6C	0.076	0.083	0.094	0.095
6E-6C	0.088	0.090	0.095	0.095
6E-6D	0.090	0.092	0.092	0.093