Revealing and understanding human protein interactome through high-throughput sequencing and meta-analysis

Proteins play a central role in human cell activities through their interactions with proteins themselves and RNAs. Yet we lack technologies that can profile protein-protein interactions (PPIs) and RNA-protein interactions (RPIs) both effectively and efficiently. For the existing technologies, the search space of proteins is limited, and the up-scaling of the products is resource-intensive. Meanwhile, the variability of PPIs detected from different technologies also leads to the lack of consensus on the architecture of the profiled PPI networks. This dissertation work presents PROPER-seq (protein-protein interaction sequencing) and PRIM-seq (protein-RNA interaction mapping by sequencing), two time-effective technologies to map cell-wide protein-protein interactions (PPIs) and RNA-protein interactions (RPIs) respectively in vitro. This dissertation work also utilizes PPIs derived from PROPER-seq to induct human PPI network features. The technologies and analysis together provide rich resources to the community for studying protein-related interactions and understanding human proteome.In Chapter 1, I describe PROPER-seq to map PPIs. I showed the PROPER-seq identified PPIs are of robust reproducibility, precision, and recall performance. I present PROPER v1.0, a human PPI network that consists of 8,635 proteins and 210,518 interactions. I delivered PROPERseqTools and PROPER v1.0 database, an open-source software, and an online database for people to process PROPER-seq libraries and to better access PROPER v1.0. In Chapter 2, I describe my analysis of utilizing multiple human PPI datasets to systematically examine the architectural characteristics of human PPI networks. I found consistent evidence to support that a comprehensive human PPI network should be a scale-free network filled with many completed or close-to-completed cliques. The hub proteins with similar molecular functions are often highly inter-connected in cliques and serve as building blocks in small network motifs. In Chapter 3, I describe PRIM-seq to systematically map RNA-protein interactions (RPIs) in vitro. I present PRIM v.1.0, a human RPI network that consists of 117,516 RPIs from 8,440 RNAs and 7,691 proteins. I showed the enrichment of previously characterized RNA-binding proteins (RBPs) in PRIM v.1.0 and found evidence to support PHGDH as an RBP