SST-editing

SST-editing enables gene expression-guided editing of immunofluorescence images for spatial transcriptomic data using GAN inversion techniques to model cellular state transitions between normal and tumor tissue samples.

Key Features:

• In Silico Spatial Transcriptomic Editing: Performs gene expression-guided editing of immunofluorescence images at single-cell resolution.
• GAN Inversion Framework: Trains models using cell-level spatial transcriptomics data extracted from normal and tumor tissue slides under the framework of GAN Inversion.
• Cellular State Transition Simulation: Simulates cellular state transitions by feeding edited gene expression levels to trained models.
• Tumor-to-Normal Transition Modeling: Models the transition from tumor to normal tissue samples, quantified using interpretable cellular features compared to normal cellular images as ground truth.

Scientific Applications:

• Spatial Transcriptomics Image Editing: Enables editing of immunofluorescence images guided by gene expression data for spatial transcriptomics technologies with matched gene expression and biomedical image data.
• Cancer Tissue State Modeling: Simulates and models transitions between normal and tumor cellular states in tissue samples.

Methodology:

SST-editing extracts cell-level spatial transcriptomics data from normal and tumor tissue slides and trains the approach under the framework of GAN Inversion. Edited gene expression levels are fed to trained models to simulate cellular state transitions.