Fig. 3.

Protein LFQ by weighted geometric means of peptide abundances. The inter-quantile ranges (boxes) and 90% percentile range (whiskers) of relative quantifications of 2800 informative human proteins (left panel) and 1564 yeast proteins (right panel) in 20 mixtures. Concentration unit: ng per 6 μl. Relative quantification using weighted geometric mean showed tight linear relation to the actual protein concentration in the mixtures. Human proteins appeared to have smaller quantification errors compared with those of the yeast proteins, because of an on average larger number of constituent peptides per protein. Results of three other summarizing approaches are presented in supplemental Fig. S3.