HIPPIE: Integrating Protein Interaction Networks with Experiment Based Quality Scores

Abstract

Protein function is often modulated by protein-protein interactions (PPIs) and therefore defining the partners of a protein helps to understand its activity. PPIs can be detected through different experimental approaches and are collected in several expert curated databases. These databases are used by researchers interested in examining detailed information on particular proteins. In many analyses the reliability of the characterization of the interactions becomes important and it might be necessary to select sets of PPIs of different confidence levels. To this goal, we generated HIPPIE (Human Integrated Protein-Protein Interaction rEference), a human PPI dataset with a normalized scoring scheme that integrates multiple experimental PPI datasets. HIPPIE's scoring scheme has been optimized by human experts and a computer algorithm to reflect the amount and quality of evidence for a given PPI and we show that these scores correlate to the quality of the experimental characterization. The HIPPIE web tool (available at http://cbdm.mdc-berlin.de/tools/hippie) allows researchers to do network analyses focused on likely true PPI sets by generating subnetworks around proteins of interest at a specified confidence level.

Are protein-protein interfaces special regions on a protein’s surface?

“…The HIPPIE 53,54 provided the source of experimental interactions. The HIPPIE data set is comprised of ∼200 000 interactions made by ∼15 000 human proteins, providing 13 interactions per protein on average.…”

ProfPPIdb: pairs of physical protein-protein interactions predicted for entire proteomes

“…However, none of those sequence-based 27 methods restrict their method to the identification of physical non-permanent PPIs as 28 we defined them. Most of those methods used permanent complexes, the others also 29 associations. This is also true for methods pioneering the use of kernel-based 30 predictions [14,15].…”

“…However, we 76 also need to remove redundancy in many non-trivial ways [23]. We used an established 77 expert knowledge-scoring scheme [29] to reflect the quality of evidence for a given PPI. 78 The scheme assigned scores from one (lowest reliability) to ten (highest reliability) for 79 each experimental method used to annotate a PPI.…”

“…We also found evidences of PPIs for PPIs which our models did not predict any 215 direct physical interaction. However, these evidences were usually experimental 216 evidences with expert knowledge scores of lower or equal 4 [29] and thus highly likely to 217 be false positives. A more detailed description of our findings can be found in the 218 supplementary materials Appendix A.2.…”

Genome-wide approaches and technologies to assess human variation

“…For exome 53 sequencing, or other targeted approaches, target capture and enrichment is implemented prior 54 to amplification. Prior to massively parallel sequencing, human genomic sequencing already 55 used target enrichment approaches, but these were laborious and not highly scalable methods 56 such as PCR for specific segment amplification or cloning of discreet genomic segments using 57 bacterial vectors and including fosmid and BAC library construction. The currently used targeted 58 capture methodologies were originally developed using oligonucleotide probes covalently 59 bound to a solid array glass slide designed to specifically bind target regions or the exons of the 60 target genes [2][3][4][5].…”