SpaTRACE: Spatiotemporal recurrent auto-encoder for reconstructing signaling and regulatory networks from...

Area of Science:

• Computational Biology
• Systems Biology
• Genomics

Background:

• Cell-cell communication and gene regulatory programs (GRNs) are crucial for coordinating cellular behaviors.
• Spatial transcriptomics offers insights into gene expression within spatial contexts but often analyzes signaling or GRNs separately.
• Existing methods have limitations in capturing temporal dynamics and discovering novel interactions due to reliance on curated databases and steady-state assumptions.

Purpose of the Study:

• To develop a novel computational framework, SpaTRACE, for the joint inference of intercellular signaling and gene regulatory networks from spatial transcriptomics data.
• To model time-lagged dependencies and capture dynamic regulatory processes along cellular trajectories.
• To enable the discovery of novel signaling pathways and regulatory interactions without relying on predefined databases.

Main Methods:

• SpaTRACE utilizes a spatiotemporal recurrent autoencoder with an attention-based encoder-decoder architecture.
• It models time-lagged dependencies along pseudotime-sampled cellular trajectories.
• The framework predicts future gene expression from upstream intra- and intercellular signals, simultaneously reconstructing TF-TG, LR-TG, and LR binding interactions.

Main Results:

• SpaTRACE accurately recovers both GRN and signaling interactions in synthetic benchmarks, outperforming existing methods.
• The framework successfully identified transcriptional regulators and signaling programs associated with neuronal differentiation in mouse midbrain development.
• Analysis of axolotl brain regeneration revealed stage-specific signaling dynamics and candidate interactions crucial for tissue repair.

Conclusions:

• SpaTRACE provides a powerful, integrated approach for dissecting complex spatiotemporal regulatory mechanisms from spatial transcriptomics data.
• The method advances the study of dynamic cell-cell communication and gene regulation in development, regeneration, and disease.
• SpaTRACE facilitates the discovery of novel biological interactions and pathways, enhancing our understanding of cellular coordination.