Fig. 6: Clinical colorectal cancer urinary proteomics analysis among three LC-MS platforms.

a Overlap of identified protein groups from Lumos, E480, and TIMS. b, c The unsupervised learning t-Distributed Stochastic Neighbor Embedding (t-SNE) plot overview of urinary proteomics among Lumos, E480, and TIMS (b) and colorectal cancer (CRC) and healthy control (HC) group. d The number of proteins that significantly different (q-value < 0.05) in abundance by CRC and HC within each platform. e-g Correlation of protein CRC/HC fold changes in pairwise combinations of three platforms. Combinations are E480 vs. Lumos (e), E480 vs. TIMS (f), Lumos vs. TIMS (g). h Heatmap of the dysregulated canonical pathways between CRC and HC in the three platforms depicted by IPA ingenuity pathway analysis. Red: Z_score > 0, activated; Blue: Z score <0, inhibited. i Receiver operating characteristic (ROC) curve for the Support Vector Machines (SVM)-based model to classify CRC vs. HC individuals when trained on E480 data. j–l Confusion matrix showed the model performance for classifying CRC vs. HC individuals. Source data are provided as a Source Data file.