TRANSPIRE

TRANSPIRE identifies intracellular protein translocation events from spatial proteomics data to elucidate dynamic changes in protein subcellular localization and their regulatory and pathway-level implications.

Key Features:

• Probabilistic Gaussian Process Classifier: Employs a probabilistic Gaussian process classifier trained on synthetic translocation profiles derived from organelle marker proteins to predict changes in protein distribution across cellular compartments.
• Integration with Co-translocating Proteins and Gene Ontology Associations: Integrates predicted translocations with co-translocating proteins and complexes and annotates results with Gene Ontology associations for functional interpretation.
• Validation and Application in Viral Infection Studies: Validated for predicting nuclear–cytoplasmic shuttling and applied to Kaposi Sarcoma-associated herpesvirus (KSHV) mRNA decay and human cytomegalovirus (HCMV) infection data, including prediction of over 800 protein translocations during HCMV replication.
• Biological Insights from Translocation Events: Links translocation events to host defense, metabolism, cellular trafficking, and Wnt signaling, reporting specific movements such as LDLR translocation to the lysosome (validated by targeted mass spectrometry) and DAPK3 kinase relocalization (validated by microscopy).

Scientific Applications:

• Viral infection proteome remodeling: Characterizes virus-induced organelle remodeling and widespread localization changes during KSHV and HCMV infection.
• Nuclear–cytoplasmic shuttling analysis: Detects nuclear–cytoplasmic translocations of RNA-binding proteins and other factors involved in mRNA metabolism.
• Pathway-level interpretation: Connects localization changes to Gene Ontology terms to infer impacts on host defense, metabolism, cellular trafficking, and Wnt signaling.
• Target prioritization for validation: Prioritizes proteins for targeted mass spectrometry and microscopy-based validation, exemplified by LDLR and DAPK3.

Methodology:

A probabilistic Gaussian process classifier is trained on synthetic translocation profiles derived from organelle marker proteins and predictions are integrated with co-translocating protein information and Gene Ontology associations.