Proteome-Scale Human Interactomics

Abstract

Cellular functions are mediated by complex interactome networks of physical, biochemical, and functional interactions between DNA sequences, RNA molecules, proteins, lipids, and small metabolites. A thorough understanding of cellular organization requires accurate and relatively complete models of interactome networks at proteome scale. The recent publication of four human protein-protein interaction (PPI) maps represents a technological breakthrough and an unprecedented resource for the scientific community, heralding a new era of proteome-scale human interactomics. Our knowledge gained from these and complementary studies provides fresh insights into the opportunities and challenges when analyzing systematically generated interactome data, defines a clear roadmap towards the generation of a first reference interactome, and reveals new perspectives on the organization of cellular life.

Figure 1. Published systematic human protein interactome maps

It took about ten years from the first publication of a medium-scale set of human PPIs to the publication of several proteome-scale systematic maps of biophysical relationships between pairs of human proteins (BPPs). The color code indicates the primary screening method used. The displayed datasets are, from left to right, Stelzl et al. [58], Rual et al. [59], Ewing et al. [60], CoFrac-12 [17], Kristensen et al. [61], HI-II-14 [5], BioPlex [4], CoFrac-15 [6], and QUBIC [3].

Figure 2. Overview of the generation of four recently published human interactome maps

Empty circles represent expression constructs, filled circles represent proteins. NGS = next-generation sequencing, DB = DNA-binding domain and AD = activation domain of the Gal4 transcription factor. The number of pairs and proteins are determined after mapping each dataset to the Entrez Gene ID gene space. The maps are HI-II-14 [5], BioPlex [4], QUBIC [3], and CoFrac-15 [6].

Figure 3. Genome-wide coverage biases of functional and biophysical protein networks

Interactome maps are each represented as an adjacency matrix in which all genes in the human genome were ranked based on publication count as extracted from [5] and grouped into bins of 478. The color scale was adjusted for every network to range from 0 to the largest number of protein pairs observed in any bin-by-bin subspace. Sources of the networks: Lit-BM-13 extracted from [5], HI-I-II merge of [5, 59], BioPlex [4], GO (Gene Ontology) network extracted from [5] (briefly, pairs of proteins were built if they share GO annotations, GO terms were filtered to those with at most 30 annotated genes), QUBIC [3], CoFrac-12-15 merge of [6, 17]. Every dataset was mapped to the Entrez Gene ID gene space. Observed study biases in the BioPlex, QUBIC and CoFrac-12-15 datasets are likely related to their propensity to detect interactions between more highly expressed genes.

Figure 4. Overlaps between the four human interactome maps

All of the displayed datasets (same as in Figure 3) were mapped to the Entrez Gene ID gene space prior to calculating their overlaps. The interactome maps were not restricted to the common gene space that was screened by all four studies.

Figure 5. Comparison of the different quality control approaches for generating the four human interactome maps

HA = hemagglutinin, GFP = green fluorescent protein, HumanNet [62], CORUM [63], STRING [64], GeneMania [65], RAB11B = Ras-related protein Rab-11B, FDR = false discovery rate (fraction of all identified BPPs that are false positives), sensitivity (fraction of “real” BPPs identified), FPR = false positive rate (fraction of negative BPPs scored as positive), precision (fraction of all reported BPPs that are correct), MAPPIT = mammalian protein-protein interaction trap [11], PCA = protein complementation assay [10], wNAPPA = well nucleic acid programmable protein array [66], coIP = co-immunoprecipitation, PAM-SILAC = purify after mixing stable isotope labeling of amino acids in cell culture.