Poster #20 - Milad Imeni Markhali

Beyond Covariance: Utilizing Spatial Transcriptomics to Reveal Local Gene Expression and Co-Expression Patterns in a Preclinical Model of TMJ Dysfunction.

Milad Imeni Markhali, PhD Student, Center for Bioinformatics and Computational Biology, University of Delaware Austin Keeler, PhD, Department of Biological Sciences, University of Delaware; Center for Bioinformatics and Computational Biology, University of Delaware Joohyun Lim, PhD, Department of Biological Sciences, University of Delaware; Center for Bioinformatics and Computational Biology, University of Delaware

Loss-of-function mutations in FKBP10 cause skeletal dysplasia and joint abnormalities in human patients. Previous studies have shown that Fkbp10 is essential for collagen cross-linking in fibrous tissues such as bone and tendon. In addition, tissue-specific removal of Fkbp10 induces ectopic chondrogenesis and progressive ankle joint deformities in postnatal mice. Unexpectedly, conditional deletion of Fkbp10 resulted in accelerated bone formation and ectopic ossification in the temporomandibular joint (TMJ), which is essential for masticatory function. To investigate pathological mechanisms of TMJ dysfunction, we used a single-cell spatial transcriptomics platform (Xenium, 10X Genomics) for spatially resolved gene expression profiles across tissue regions of interest, enabling identification of key biomarkers involved in these pathological changes. Our analysis examines how spatial cell coordinates enhance understanding of Fkbp10 deletion effects. Typical workflows for spatial transcriptomics analysis start with PCA-based dimensionality reduction, followed by clustering to group cells based on similarities in their expression profiles, and then differential expression analysis to identify biomarkers, and enrichment-based annotation using these biomarkers; spatial information is then incorporated into downstream analyses. By integrating spatial context and cell neighborhood relationships, we detected location-linked gene expression patterns in tissue space, constructed proximity graphs of cell clusters, and fitted geometric borders between adjacent clusters using extracted 'border cells'. The central question is how spatial analysis complements post-clustering interpretation, and how combining expression profiles with local context provides insights into gene expression and co-expression patterns, beyond what is captured by covariance estimation in high-dimensional gene expression data. This was specifically explored using a selected subset of genes identified as key pathological biomarkers, with analysis of gene expression profiles tested with and without explicit spatial coordinates.