An Image-Based Data-Driven Analysis of Metabolite Clusters Architecture in Heterogeneous Colorectal Cancer Tissues
Reveal functional metabolomic heterogeneity in MSI samples with Uniform Manifold Approximation and Projection (UMAP).
Label free Mass Spectrometry Imaging (MSI) has been developed and used by ImaBiotech to detect and quantify metabolites in tissue for almost a decade. Although MSI provides a very rich information on the spatial distribution of biomolecular species in tissue section, the biggest challenge is to make sense of the large complex data set. Effective data visualization plays a role in exploring patterns/features in such big datasets. Such insights are essential to facilitate data interpretation though context-guided visualization and help develop hypotheses on newly founded biological patterns.
Clustering of metabolites within complex heterogeneous tissue
To overcome the challenge of a big and complex dataset, ImaBiotech’s R&D team has developed a visualization workflow to overlay clustered metabolic patterns on tissue in order to identify system biological models for disease progression or therapy response assessment.
This case study presents a workflow to project high dimensional spatial metabolic data on two-dimensional colon cancer tissue sections with different pathological stages of cancer.
This workflow allows to cluster metabolic components according to their biological function in the complexity of the tissue and can be applied to identify new metabolic signatures that correlate with disease prognostic or treatment prediction.
Clustering reveals differential metabolite pattern distributions according to disease progression
The clustering patterns were different across the different pathological stages of the disease and the complexity of the metabolite patterns increased with disease extent. This shows that metabolite distribution is affected during disease progression and altered by the extent of the disease .
This case study presents a visualization tool to obtain insights regarding the heterogeneity of molecular patterns in this instance metabolites present in the tissue. This will help understand the underlying biological phenomena at play in the tissue.
The approach was illustrated on heterogeneous colorectal cancer tissue and the identified metabolites signatures were shown to be differentially expressed along disease progression. This tool is a powerful tool in the search of disease mechanism or facilitate biomarker discovery.