Poster Session I. - A: Molecular Medicine
Kerestély Márk
Semmelweis University, Department of Molecular Biology
Márk Kerestély1, István Narozsny1, Réka Czinege1, Dániel Veres2, Péter Csermely1
1: Semmelweis University, Department of Molecular Biology
2: Turbine Ltd.
Introduction: For a deeper understanding of eukaryotic cell function, it is essential to map the localization of proteins and their translocations between subcellular compartments. Previously, our research group created the ComPPI database, which integrates compartmentalized protein-protein interactions, and the Translocatome database, which contains data and predictions on protein translocations. Compartmentalization data is highly useful for logical network models (like Boolean models) that can cost-effectively model complex biological systems.
Aims: Our goal was to update ComPPI and Translocatome and make them suitable for utilization in logical modelling. We also aimed to validate the previous predictions of Translocatome.
Method: We utilized the “The Minimum Information about a Molecular Interaction CAusal Statement” (MI2CAST) guidelines and the MITAB data format in the update of ComPPI and Translocatome. For the validation of Translocatome’s predictions, we manually curated translocating and non-translocating proteins and described the performance of these predictions with measures of diagnostic accuracy (e.g. ROC AUC, F1-score, MCC).
Results: We updated ComPPI and Translocatome based on the MI2CAST guidelines and integrated data into them about post-translational modifications, biological functions and protein complexes. We collected and validated 402 translocating proteins from the Reactome, UniProt and Signor databases and 162 non-translocating proteins from the UniProt, ComPPI and HPA databases. Based on these curated proteins, the statistical analysis of previous predictions of Translocatome resulted in ROC AUC = 0,73, F1 = 0,68, MCC = 0,37.
Conclusion: The update of ComPPI and Translocatome facilitates their use in logical modelling of complex biological systems. The Translocatome prediction model performed reasonably well. The curated data can be used to further train the model.
Funding: Supported by the Thematic Excellence Program (Tématerületi Kiválósági Program TKP2021-EGA-24) of the Ministry for Innovation and Technology in Hungary, within the framework of the Molecular Biology thematic program of the Semmelweis University and by the 2024–2.1.1-EKÖP-2024–00004 University Research Scholarship Programme of the Ministry for Culture and Innovation from the source of the National Research, Development and Innovation Fund.