Sign in Subscribe
Genetic research

Predicting Gene Expression Responses to Environment in Arabidopsis Thaliana Using Natural Variation in DNA Sequence

Strategic Insights

— 1 min read

Research on the evolution of gene expression responses has been ongoing, focusing on understanding how DNA sequence influences expression in response to variable environments. A recent study from Pennsylvania State University has shed light on this complex process, suggesting that flexible machine learning models can learn the underlying regulatory genotype to phenotype map. The study utilized cold-responsive transcriptome profiles in 5 diverse accessions to test this approach. Results showed that convolutional neural networks (CNNs) predicted differential expression with moderate accuracy, but were hindered by the biological complexity of regulation and the large potential regulatory code.

Key Takeaways:

  • The evolution of gene expression responses is a critical component of adaptation to variable environments.
  • Predicting how DNA sequence influences expression is challenging due to the genotype to phenotype map not being well resolved for regulatory elements, transcription factor binding, regulatory interactions, and epigenetic features.
  • A study from Pennsylvania State University tested if flexible machine learning models could learn the underlying regulatory genotype to phenotype map using cold-responsive transcriptome profiles in 5 diverse accessions.
  • 14 and 15 motifs were significantly enriched within the up- and downstream regions of cold-responsive differentially regulated genes (DEGs).
  • CNNs predicted differential expression with moderate accuracy, but were hindered by biological complexity and the large potential regulatory code.
  • Approaches to predict DEGs between specific environments based only on proximate DNA sequences require further development and additional information may be required.
  • The study was conducted by researchers from Pennsylvania State University, including Emily S. Bellis, Margarita Takou, and Jesse R. Lasky.

Statistics:

  • 5 diverse accessions were used in the study to test the approach.
  • 14 and 15 motifs were significantly enriched within the up- and downstream regions of cold-responsive DEGs.
  • CNNs predicted differential expression with moderate accuracy.
  • 2737 is the pagination number for the published study in Science Letter.

Sources:

  • bioRxiv, 2025
  • Science Letter, October 17, 2025; p 2737.

Read more