Proteonano™ enables deep proteomic analysis of plasma samples from a community cohort with different levels of cognitive defects
Figure 1: Sample preprocessing and LC-MS analysis.
4347 protein groups were identified in 206 plsama samples. 2704 of them were mapped to the human plasma protein project protein catalog. Concentrations of these proteins spanned 9 orders of magnitude, with lowest protein concentration of 1.6 pg/mL.
Differentially expressed proteins between patient groups
Figure 2. Differentially expressed proteins (DEP) in different patient groups.After data normalization, DEPs were determined. (A) 64 DEPs between N and MCI groups.
Differential protein expression analyses showed 64, 91, and 159 proteins had different abundance between MCI and N, D and MCI, and D and N groups, respectively. The most upregulated protein in MCI group relative to N group was non-erythrocytic β spectrin (SPTBN1) and most downregulated protein was matrix Gla protein (MGP).
Identification of effective biomarker combinations to differentiate normal and MCI patients
Figure 3. Assessment of multivariate models.
Multivariate analyses were subsequently carried out. Eight f eature selection methods were employed, including least absolute shrinkage and selection operator (LASSO) and random forest (RF). The best model diff erentiating MCI and N groups contains nine proteins, with a ROC-AUC of 0.92 (5-95 % confidence interval:0.89-0.98), and PR-AUC of 0.84.
