IonSpattern

IonSpattern implements unsupervised spatial segmentation and characterization of ions in mass spectrometry imaging (MSI) data to identify morphology-specific and single-ion spatial distributions within biological tissues.

Key Features:

• Spatial-Dirichlet Gaussian Mixture Model (Spatial-DGMM): Implements the Spatial-DGMM algorithm to detect components within single-ion images and incorporate spatial structure specific to MSI data.
• Unsupervised Segmentation: Performs segmentation of ion images without labeled training data to delineate spatially coherent ion components.
• Morphology-Specific Analysis: Evaluates ion localization and regulation across tissue morphologies and distinguishes homogeneous versus heterogeneous spatial ion distributions.
• Downstream Analyses Support: Extracts spatial features for detecting groups of ions with similar spatial patterns and for assessing changes in tissue chemical composition under different conditions.

Scientific Applications:

• Disease Pathology: Maps morphology-specific ion distributions to investigate disease-related biochemical changes in tissues.
• Drug Delivery Mechanisms: Characterizes spatial distribution of drug-related ions to study delivery and localization within tissues.
• Tissue Engineering: Assesses chemical heterogeneity and ion localization to inform tissue engineering and regenerative studies.
• Comparative Spatial Biochemical Analysis: Enables comparison of spatial chemical composition across conditions or treatments in MSI experiments.

Methodology:

The Spatial-DGMM algorithm is executed via R scripts W_matrix.R, GMM.R, k_DGMM.R and S_DGMM.R, and the script CpG_Sal_mouse_brain.R provides an example analysis for CpG and Sal preconditioned mouse brain data.