Dataset Information

The genetic architecture of protein stability

ABSTRACT: There are more ways to synthesize a 100 amino acid protein (20^100) than atoms in the universe. Only a miniscule fraction of such a vast sequence space can ever be experimentally or computationally surveyed. Deep neural networks are increasingly being used to navigate high-dimensional sequence spaces. However, these models are extremely complicated and provide little insight into the fundamental genetic architecture of proteins. Here, by experimentally exploring sequence spaces >10^10, we show that the genetic architecture of at least some proteins is remarkably simple, allowing accurate genetic prediction in high-dimensional sequence spaces with fully interpretable biophysical models. These models capture the non-linear relationships between free energies and phenotypes but otherwise consist of additive free energy changes with a small contribution from pairwise energetic couplings. These energetic couplings are sparse and caused by structural contacts and backbone propagations. Our results suggest that artificial intelligence models may be vastly more complicated than the proteins that they are modeling and that protein genetics is actually both simple and intelligible.

ORGANISM(S): Saccharomyces cerevisiae

PROVIDER: GSE246322 | GEO | 2023/10/27

REPOSITORIES: GEO

ACCESS DATA

Dataset's files

Source:

			Action	DRS
		Other

Items per page:

1 - 1 of 1

Similar Datasets

Project description:The traditional method for studying cancer in vitro is to grow immortalized cancer cells in two-dimensional (2D) monolayers on plastic. However, many cellular features are impaired in these unnatural conditions and big alterations in gene expression in comparison to tumors have been reported. Three-dimensional (3D) cell culture models have become increasingly popular and are suggested to be better models than 2D monolayers due to improved cell-to-cell contacts and structures that resemble in vivo architecture. The aim of this study was to develop a simple high-throughput 3D drug screening method and to compare drug responses in JIMT1 breast cancer cells when grown in 2D, in polyHEMA coated anchorage independent 3D models and in Matrigel on-top 3D cell culture models. We screened 102 compounds with multiple concentrations and biological replicates for their effects on cell proliferation. The cells were either treated immediately upon plating or they were allowed to grow in 3D for four days prior to the drug treatment. Big variations in drug responses were observed between the models indicating that comparisons of culture model influenced drug sensitivities cannot be made based on effects of a single drug. However, we show with the 63 most prominent drugs that, in general, JIMT1 cells grown on Matrigel were significantly more sensitive to drugs than cells grown in 2D cultures, while responses of cells grown in polyHEMA resembled those of 2D. Furthermore, comparison of gene expression profiles of the cell culture models to xenograft tumors indicated that cells cultured in Matrigel and as xenografts most closely resembled each other. In this study we also suggest that 3D cultures can provide a platform for systematic experimentation of larger compound collections in a high-throughput mode and be used as alternatives for traditional 2D screens towards better comparability to in vivo state. Gene expression analysis of JIMT1 breast cancer cells cultured as xenografts for 43 days, in two dimensional cultures for seven days (2D7d), in polyHEMA three dimensional cell culture models for four and seven days (PH7d and PH7d), and in Matrigel three dimensional cultures for four and seven days (MG4d and MG7d). Two biological replicates was included for each sample.

Dataset Information

The genetic architecture of protein stability

Dataset's files

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets