Functional Annotation and Motif Discovery Leveraging Protein Interaction Networks

Abstract

Proteins play essential roles in a variety of cellular processes. To accomplish their functions, proteins often form stable or transient interactions, resulting in a complex network of protein-protein interactions (PPIs). These networks are rich in biological information regarding the molecular processes underpinning cellular biology and function. In this thesis, I explore methods to leverage PPI networks to infer novel biological insights. Functional enrichment analyses aim to identify the over-represented biological functions among a list of genes of interest, typically derived from high-throughput experiments measuring gene or protein expression. When analyzing multiple conditions, these methods often rely solely on the differential expression of individual genes or proteins, overlooking their connectivity within PPI networks and the functional relationships that can strengthen biological insights. To address this limitation, I developed PIGNON, a novel graph theory-based method for functional enrichment analysis. PIGNON combines large-scale PPI networks with protein differential expression to uncover functional annotations associated with proteins that are both differentially expressed and significantly clustered in the PPI network. When applied to a quantitative proteomics analysis of different molecular subtypes of breast cancer, PIGNON detects both well-characterized and novel biological processes associated with these subtypes. Beyond shared functionality, PPI networks can be used to investigate other factors that may influence protein regulation. A key mechanism involves mRNA 3'UnTranslated Regions (3'UTRs) sequence and structural elements. 3'UTRs are downstream of the coding sequence in the mRNA transcript of genes. They are known to modulate mRNA transcript and protein localization by recruiting specific binding factors. Advances in proximity labelling coupled with mass-spectrometry (e.g. BioID-MS) have expanded our ability to capture proximal protein relationships within cells, generating protein co-localization networks. Indeed, proteins that are densely connected in such networks are likely to be located in similar cellular compartments. If such proteins share a common 3'UTR sequence or structural motif in their associated mRNAs, the motif could be directly or indirectly related to their localization. Therefore, I developed frameworks for the discovery of both sequence and structural motifs in 3'UTRs associated with significantly clustered proteins within protein co-localization networks. Overall, my work emphasizes the power of PPI network-based approaches to deepen our systems-level understanding of cellular function, offering new avenues for biological discovery from quantitative studies and the annotation of regulatory elements in mRNA transcripts.