Off the deep end: What can deep learning do for the gene expression field?

Raicu, Ana‐Maria; Fay, Justin C.; Rohner, Nicolas; Zeitlinger, Julia; Arnosti, David N.

doi:10.1016/j.jbc.2022.102760

Cited by 1 publication

(1 citation statement)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Using machine learning methods in fields such as genomics and population genetics can help navigate increasingly bigger datasets and reveal complex patterns (Korfmann et al 2023). Especially in gene regulation and evolution, deep learning approaches can have an advantage over traditional methods in decoding enhancer grammar of gene regulation, as these models can learn complex cis -regulatory rules in a precise manner not biased by current knowledge (Raicu et al 2023) and can outperform more traditional methods, like clustering by kmer or linear regression in predicting gene expression (Chen et al 2016). For example, training a machine learning model on the imputed cis haplotypic information of RNA expression, Giri et al (2021) managed to exclude the impact of trans effects on the gene expression variation.…”

Section: Introductionmentioning

confidence: 99%

Predicting gene expression responses to environment inArabidopsis thalianausing natural variation in DNA sequence

Takou,

Bellis,

Lasky

2024

Preprint

View full text Add to dashboard Cite

The evolution of gene expression responses are a critical component of adaptation to variable environments. Predicting how DNA sequence influences expression is challenging because the genotype to phenotype map is not well resolved for cis regulatory elements, transcription factor binding, regulatory interactions, and epigenetic features, not to mention how these factors respond to environment. We tested if flexible machine learning models could learn some of the underlying cis-regulatory genotype to phenotype map. We tested this approach using cold-responsive transcriptome profiles in 5 diverse Arabidopsis thaliana accessions. We first tested for evidence that cis regulation plays a role in environmental response, finding 14 and 15 motifs that were significantly enriched within the up- and down-stream regions of cold-responsive differentially regulated genes (DEGs). We next applied convolutional neural networks (CNNs), which learn de novo cis-regulatory motifs in DNA sequences to predict expression response to environment. We found that CNNs predicted differential expression with moderate accuracy, with evidence that predictions were hindered by biological complexity of regulation and the large potential regulatory code. Overall, DEGs between specific environments can be predicted based on variation in cis-regulatory sequences, although more information needs to be incorporated and better models may be required.

show abstract