Simple patterns found in complex microbial communities

Researchers discovered that the changing dynamics of observable microbes can be captured by low-dimensional representations.

Microbial communities often show complex dynamics. These are driven by many interactions among community members and their environments.

The temporal dynamics of observable microbial populations can be captured by low-dimensional representations. This reveals an emergent simplicity.

Researchers used variational autoencoders to define a critical latent dimension, EcE_c. This EcE_c quantifies the smallest latent dimension needed for high-fidelity reconstruction of observable microbial dynamics.

The EcE_c scales linearly with the number of observables. This provides an empirical principle for determining representation complexity.

This principle holds across simulations of pairwise interacting communities and consumer-resource dynamics. It also works for spatially distributed microbial patches and plasmid transfer.

The findings hold for experiments with engineered and environment-derived communities and human microbiomes. Low-dimensional representations retain information useful for selected prediction tasks.

Key facts

  • The critical latent dimension EcE_c scales linearly with the number of observables.
  • Low-dimensional representations can capture observable microbial dynamics despite complex unobserved background dynamics.
  • The scaling law holds for human microbiomes and engineered communities.

All sections

Edition