COVIDpro

COVIDpro aggregates a curated repository of proteomic expression data from COVID-19 patients to support analysis of protein dysregulation and biomarker discovery.

Key Features:

• Extensive Data Repository: Aggregates proteomic expression data from 41 original studies across 32 hospitals, comprising 3,077 patients, 5,434 samples, 14,403 unique proteins, and 19 clinical specimen types including plasma and sera.
• Analysis and Visualization Tools: Provides integrated analysis and visualization for meta-analysis, differential expression across disease subgroups, and visualization of pathways and protein interactions.
• Predictive Biomarker Identification: Identifies circulating proteins that differentiate severe from non-severe cases, nominating candidates such as lipopolysaccharide-binding protein which is highly expressed in severe disease blood samples.
• Proteomic Signatures for Classification: Enables classification of severe COVID-19 based on proteomic signatures with reported mean AUC of 0.87 and accuracy (ACC) of 0.80 across five test sets.
• Cross-cohort Integration: Integrates datasets from over 3,000 patients across 11 nations to support identification and verification of protein biomarkers and therapeutic targets.

Scientific Applications:

• Biomarker Discovery and Validation: Supports identification and verification of circulating protein biomarkers and nominates candidates such as lipopolysaccharide-binding protein for severity prediction.
• Classification and Prediction: Facilitates development and assessment of proteomic classifiers to distinguish severe versus non-severe COVID-19 (mean AUC 0.87, ACC 0.80).
• Comparative Proteomics Across Specimens: Enables comparison of protein expression across 19 clinical specimen types, including plasma and sera, to investigate tissue- and specimen-specific signatures.
• Meta-analysis Across Cohorts: Enables cross-study meta-analysis across 41 studies and multiple hospitals to evaluate reproducibility of protein dysregulation.
• Therapeutic Target Prioritization: Supports nomination and verification of dysregulated proteins as potential therapeutic targets.

Methodology:

Expert curation of proteomic data from publications up to May 2022 sourced from ProteomeXchange and PubMed, aggregated into a comprehensive dataset for comparative analysis of protein expression across disease subgroups.