Computational frameworks for predicting protein interactions via single-cell proximity sequencing

Proximity sequencing (Prox-seq) is a method that can simultaneously measure gene expression, protein abundance and protein complex information in single cells. By making use of proximity information from proteins, Prox-seq can infer surface protein complexes at the single-cell level. Previously, we proposed a statistical method for predicting protein complex abundance in single cells from Prox-seq data. However, it is difficult to experimentally validate the prediction results. Here, we first introduces a physical model which enables simulation of Prox-seq data. The simulated data is then used to validate the statistical method that we proposed previously, and to understand how protein abundance can influence protein complex prediction. Second, we propose a modification to Prox-seq, and use the new measurements from the modification to develop a second, independent method for predicting protein complexes. We demonstrate that on experimental data, the two methods generally agree with each other, with the second method generally producing more accurate protein complex abundance results. Thus, we present a simple way to investigate the behavior of Prox-seq data under various biological conditions, and independent methods for predicting and quantifying protein complexes. Prox-seq analysis of single Jurkat and Raji cells using plate-based pipeline