leptomycin b Workflow Mascot Distiller Mascot Skyline

Workflow 8: Mascot Distiller/Mascot/Skyline/MS Signal analysis. Peaklist creation was performed with Mascot Distiller as described in workflow 3, then database searches were performed with Mascot and validated with Scaffold as described for workflow 4. XIC signal corresponding to all validated peptides were extracted using the Skyline software (Skyline version v2.5, daily updates of April 2014, https://skyline.gs.washington.edu). This method was well described by Schilling et al (Schilling et al. MCP, 2012). Total areas, corresponding to the sum of the 3 extracted leptomycin b areas, were used for statistical analysis. Statistical analysis. For pairwise comparisons of samples spiked at different concentrations of UPS1, same statistical tests and fold change criteria were applied to the quantitative data obtained from each workflow, as follows: When working on spectral count metrics (workflows 1-2-3-4), a beta-binomial test was performed based on triplicate MS/MS analyzes. p-values were calculated with the software package BetaBinomial_1.2 [14] implemented in R. Fold change was calculated as ratio of average spectral counts from both conditions. For proteins absent in all replicates of one specific condition, their spectral count values were modified by adding 1 spectrum to all 6 samples in order to be able to calculate a fold change for these particular proteins. When working on MS signal intensity-based metrics (workflows 5–6–7–8), proteins were filtered out if they were not quantified in at least all replicates from one condition. Missing protein intensity values were replaced by a constant value calculated independently for each sample as the 5-percentile value of the total population. A welch t-test (two-tailed t-test, unequal variances) based on triplicate MS analyzes was then performed on log2 transformed values using the Perseus toolbox (version 1.4.0.11; http://141.61.102.17/perseus_doku). Criteria used to classify the proteins were the Welch t-test difference calculated by Perseus (difference between the two compared conditions of the mean log2 transformed value for triplicate MS/MS analyzes), and the Welch t-test p-value. Construction of the mixed dataset and plot of the ROC curves for each workflow For each workflow, quantitative outputs from the 3 pairwise comparisons described in Fig. 2 were merged in a single Excel table. This composite table is shown for each of the 8 tested workflows in a separate sheet in Supplementary Table 1. The first column indicates the origin of the quantitative values: comparison A (50Vs0.5), comparison B (50Vs5) or comparison C (25Vs12.5). The second column indicates whether the protein originate from the background (yeast) or from the spiked mixture (UPS). These UPS1 proteins are highlighted in green, red, or yellow according to the comparison in which they were quantified (A, B, and C respectively). Following columns indicate:z-score= (Welch t-test difference)- Median [(Welch t-test difference) for all quantified proteins] /Standard deviation [(Welch t-test difference) for all quantified proteins] To classify proteins as variant and non-variant and plot ROC curves, different combinations of criteria were tested: Fig. 3 shows ROC curves (sensitivity versus FDP) which illustrate how the dataset can be useful for selecting the most efficient classification filters (Fig. 3A) or for the comparison of software tools (Fig. 3B).
Acknowledgments This work was funded through the French National Agency for Research (ANR) (Grant ANR-10-INBS-08; ProFI project, “Infrastructures Nationales en Biologie et Santé”; “Investissements d’Avenir” call).
1 Data Using human microRNA v2 panel (Illumina) we have identified 586 miR species, which were evidently expressed in MSCs (see Table 1). We selected miR-92a as a one of the most abundant angio-miRs expressed in MSCs and confirmed its expression by real-time PCR [1]. Then, we overexpressed or down-regulated its content using nucleofection. We examined viability of transfected cells, which was about 90% and did not differ between cells transfected with pre-miR-92a, anti-miR-92a or scramble oilgos (Fig. 1).