Validation of AVA‐Seq using a human reference protein‐protein interaction set

Abstract

Here is an in-depth characterization of a gold-standard human protein interaction set (HsPRS-v2) using AVA-Seq's (all-vs-all sequencing) convergently fused fragment design. Selected protein pairs were sheared into ~500 bp fragments, selected for the open reading frame, assembled into the pAVA plasmid and integrated for protein interactions. Initially, these data were interpreted strictly from a binary perspective in order to compare this method to other binary protein-protein interaction (PPI) methods which utilize the same HsPRS-v2. AVA-Seq was able to recapitulate 20 of 43 (46.5%) known binary PPIs from this positive reference set; 5 of these interactions were not captured by other methods previously reported. The use of this gold-standard set allows for a better understanding of the limitations and advantages of AVA-Seq. For example, there are PPIs which were not detected even though 95% of the amino acids between the two proteins were represented. Conversely, interactions were detected with as little as 16% protein coverage. The average protein coverage of all protein pairs used in this study was 60% while the average coverage for a detected interaction pair was 68%. By increasing the coverage of the proteins to >70%, the detection of expected PPIs in this data set is expected to increase to >50%. The same experimental data allowed for the determination of other interactions within this known interacting set. To that end, AVA-Seq has detected >500 PPIs from the HsPRS-v2 proteins tested; many of which are found with multiple unique fragment pairs, in multiple screening libraries, in multiple growth conditions and in both orientations. Ongoing work aims to make enhancements to this system for application of small, focused protein interaction sets (such as this study), as well as determining the entire interactomes, among others. The ease, swiftness, and reliability of AVA-Seq allows the user to determine PPIs in a protein pool of their choosing and also interrogate the active site regions of proteins with high-resolution by taking advantage of the convergent fragment feature of this method.