Extended Data Fig. 2: Supplement for Fig. 2.

(a) Budding yeast of different sizes were metabolically labeled in cell culture and subjected to proteomic analysis. Strategies to separate cells by size are described in Fig. 2 and Extended Data Fig. 3. See methods for processing steps prior to mass spec data acquisition. (b) After cell lysis, protein concentrations were quantified. An equal amount of protein from small-, medium-, or large-cell size populations (each possessing a unique metabolic SILAC label) were mixed. Loading error from the mixing process was normalized as described in the methods. Rather than normalize L/H and L/M SILAC ratios separately, we normalize all three channels together so that the values in our dataset represent relative changes across all cell sizes. Data normalization from a poorly mixed hypothetical example experiment is depicted. (c) For each individual peptide triplet, we determined the fraction of the triplet’s total ion intensity present in each SILAC channel. The distributions of these fractions were then adjusted by the median (see methods for a complete description of the normalization process). Data from the hypothetical example in (b) is depicted. (d) Peptide slope values are calculated from a linear regression of the relative ion intensity in each SILAC channel and mean cell size. Mean cell size was determined by Coulter counter prior to mixing and lysis. The mean squared error was used to track the linear fit of each peptide triplet regression. (e) Correlation of peptide slopes calculated from replicate experiments before and after applying a filter for mean squared error (MSE). 146,492 unique peptide measurements were identified in two replicate experiments (“G1 arrest time” from Fig. 2). A unique peptide measurement is defined by the peptide sequence, modification state, charge state, and pre-fractionation fraction number. Loosely filtering peptide triplets by MSE increased the correlation between biological replicate experiments.