Answers are grounded in this paper and cite their sources.
Plant stress responses are shaped by complex transcriptional strategies that integrate multiple signals across conditions and tissues. Although differential expression analyses have provided fundamental insights into stress-responsive genes and pathways, capturing responses across multiple conditions and tissues remains a challenge. Here, we introduce an unsupervised autoencoder-based framework to explore large-scale RNA-seq data through low-dimensional latent representations to capture global transcriptional structure. By combining latent coordinates with quantitative descriptors of intensity, plasticity, and dominance, we define continuous landscapes of transcriptional programs, enabling a unified view of stress-responsive states. Applying this framework to Solanum lycopersicum transcriptomic public datasets, we characterize the organization of transcriptional strategies across leaf and root tissues under multiple biotic and abiotic stress conditions. Although leaf and root latent spaces partially overlap, they display distinct transcriptional arrangements. Within this landscape, the high-intensity/high-plasticity (HH) region defines a regime in which differences in transcriptional organization between tissues become clearly distinguishable . Root displays a broad spectrum of moderately responsive genes and a small subset of HH genes. Leaf exhibits a greater presence of dominant genes distributed along the HH continuum, while transitional genes are mostly confined to the lower range. Using proximity-based associations in latent space, tissue-specific dynamics of resistance genes and transcription factors across leaf and root tissues were disclosed. Together, these results demonstrate that plant stress responses can be interpreted as continuous transcriptional strategies embedded in structured latent landscapes, providing a scalable and data-driven framework to investigate the global organization of plant transcriptomes across complex environmental contexts.
No takes yet. Share an insight, caveat, or question.
Dublino et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: