Multifunctionality in the human proteome: insights from MultifacetedProtDB

Multifunctional proteins represent a key level of complexity in human biology, as they participate in diverse cellular processes through multiple molecular activities, domain architectures, interaction networks, regulatory mechanisms, and post-translational modifications. Despite their biological relevance, their systematic characterization remains challenging due to fragmented information dispersed across heterogeneous databases and the absence of unified resources specifically dedicated to multifunctionality. In this thesis, I present MultiFacetedProtDB (MFDB), a curated and integrative database of human multifunctional proteins. The database combines structural, functional, and phenotypic annotations from major public resources including UniProt, Reactome, Gene Ontology (GO), RCSB Protein Data Bank (RCSB PDB), the Human Protein Atlas (HPA), Human Phenotype Ontology (HPO), MONDO, GeneCards, and others. Multifunctionality is represented through multiple perspectives, including pleiotropy, multidomain organization, functional promiscuity, moonlighting behavior, and post-translational modifications, thereby capturing the breadth of multitasking mechanisms described in the literature. Beyond curation, this work applies computational analyses to expand and explore multifunctionality. Hidden Markov Models (HMMs) were employed to identify novel candidate multifunctional proteins. Finally, a large-scale expansion of the dataset to over 10,000 proteins provided a global view of multifunctional signatures within the human proteome. Together, these results establish a comprehensive and scalable methodology for the systematic characterization of protein multifunctionality, integrating curated knowledge with computational prediction and large-scale exploratory analyses. The thesis highlights the structural, functional, and regulatory diversity of human proteins and provides a foundation for future investigations in systems biology, molecular medicine, and computational annotation of multifunctional traits.