Rxivist logo

The latent geometry of the human protein interaction network

By Gregorio Alanis-Lobato, Pablo Mier, Miguel A. Andrade-Navarro

Posted 02 Nov 2017
bioRxiv DOI: 10.1101/213165 (published DOI: 10.1093/bioinformatics/bty206)

To mine valuable information from the complex architecture of the human protein interaction network (hPIN), we require models able to describe its growth and dynamics accurately. Here, we present evidence that uncovering the latent geometry of the hPIN can ease challenging problems in systems biology. We embedded the hPIN to hyperbolic space, whose geometric properties reflect the characteristic scale invariance and strong clustering of the network. Interestingly, the inferred hyperbolic coordinates of nodes capture biologically relevant features, like protein age, function and cellular localisation. We also realised that the shorter the distance between two proteins in the embedding space, the higher their connection probability, which resulted in the prediction of plausible protein interactions. Finally, we observed that proteins can efficiently communicate with each other via a greedy routeing process, guided by the latent geometry of the hPIN. When analysed from the appropriate biological context, these efficient communication channels can be used to determine the core members of signal transduction pathways and to study how system perturbations impact their efficiency.

Download data

  • Downloaded 694 times
  • Download rankings, all-time:
    • Site-wide: 21,535 out of 94,912
    • In systems biology: 600 out of 2,425
  • Year to date:
    • Site-wide: 59,500 out of 94,912
  • Since beginning of last month:
    • Site-wide: 56,878 out of 94,912

Altmetric data

Downloads over time

Distribution of downloads per paper, site-wide


Sign up for the Rxivist weekly newsletter! (Click here for more details.)