stLearn

stLearn integrates spatial location, tissue morphology, and gene expression to identify cell types, map spatial trajectories, and detect ligand–receptor interaction hotspots in spatial transcriptomics data.

Key Features:

• Data integration and normalization: Normalizes gene expression using a distance measure based on morphological similarity and neighborhood smoothing to combine spatial, morphological, and transcriptomic signals.
• Spatial clustering and sub-clustering: Performs spatial clustering and further sub-clustering based on spatial separation to refine identification of cellular populations within tissues.
• Pseudo-Space-Time (PST) distance calculation: Computes a composite PST distance that combines physical and gene expression distances to reconstruct spatial transition gradients between transcriptional states.
• Identification of interaction hotspots: Integrates spatial information and gene expression to identify tissue regions with high ligand–receptor interaction activity and diverse cell type co-localization.

Scientific Applications:

• Cell type identification: Integrates spatial and morphological data with gene expression to distinguish distinct cell types within complex tissue environments.
• Reconstruction of cell type evolution: Reconstructs trajectories of cell-type transitions within tissues to elucidate developmental processes and disease progression.
• Mapping spatial trajectories: Models spatial transitions and gradients of cellular states, enabling analysis of processes such as cancer invasion and metastasis.
• Detection of interaction hotspots: Identifies regions with elevated ligand–receptor interactions and cell co-localization to pinpoint potential sites of cell–cell communication.

Methodology:

Three core computational steps implemented in Python: normalization using morphological similarity and neighborhood smoothing followed by spatial clustering and sub-clustering; PST distance calculation via directed minimum spanning tree optimization; and hotspot detection integrating spatial and gene expression data to identify ligand–receptor interactions and cell co-localization.