ProteomeHD

ProteomeHD constructs a co-regulation map from quantitative proteomic responses to biological perturbations to enable functional annotation and discovery within the human proteome.

Key Features:

• Quantitative dataset: Abundance changes for 10,323 human proteins measured across 294 distinct biological conditions using isotope-labeling mass spectrometry.
• Data integration: Augmentation of the core dataset with additional proteomic data from the Proteomics Identifications database.
• Machine learning: Use of the treeClust algorithm to identify functional associations between co-regulated proteins.
• Co-regulation map: Generation of a detailed map representing protein associations and regulatory networks derived from co-regulation patterns.
• Non-physical relationships: Detection of functional relationships that do not require physical interaction or colocalization, exemplified by PEX11β co-regulation with mitochondrial respiration factors indicating a peroxisome–mitochondria interface.
• Microprotein prediction: Prediction of functions for microproteins based on their co-regulation profiles.

Scientific Applications:

• Functional annotation: Assigning putative functions to uncharacterized proteins using co-regulation evidence.
• Hypothesis generation: Generating hypotheses about protein function and inter-organelle relationships, including peroxisome–mitochondria interfaces (PEX11β example).
• Pathway and network analysis: Investigating complex biological processes and regulatory pathways via co-regulation networks.
• Experimental prioritization: Prioritizing candidate proteins, including microproteins, for experimental validation.

Methodology:

Implemented as an R script that analyzes isotope-labeling mass spectrometry measurements of 10,323 human proteins across 294 biological conditions, integrates additional data from the Proteomics Identifications database, and applies the treeClust machine learning algorithm to derive functional associations and construct a co-regulation map.