Nature Communications (Sep 2021)

VEGA is an interpretable generative model for inferring biological network activity in single-cell transcriptomics

  • Lucas Seninge,
  • Ioannis Anastopoulos,
  • Hongxu Ding,
  • Joshua Stuart

DOI
https://doi.org/10.1038/s41467-021-26017-0
Journal volume & issue
Vol. 12, no. 1
pp. 1 – 9

Abstract

Read online

Developing interpretable models is a major challenge in single cell deep learning. Here we show that the VEGA variational autoencoder model, whose decoder wiring mirrors gene modules, can provide direct interpretability to the latent space further enabling the inference of biological module activity.