Linear structure unfolding : application to mouse brain in spatial transcriptomics.

Area of Science:

• Computational biology
• Molecular biology
• Genomics

Background:

• Spatial transcriptomics combines molecular biology and imaging for gene expression analysis in tissues.
• Current methods face challenges in analyzing specific tissue shapes like elongated or curved regions.
• Understanding cellular and molecular distributions within complex tissue architectures is crucial.

Purpose of the Study:

• To develop a computational method for analyzing spatial transcriptomics data in elongated or curved tissue regions.
• To enable the study of cell and transcript proportions along the principal axis of linear structures.
• To provide tools for analyzing spatial distributions within complex tissue shapes.

Main Methods:

• A novel computational approach for analyzing spatial transcriptomics data.
• Utilizes k-means clustering and salesman problem optimization to compute the centerline of elongated shapes.
• Develops a method to 'unfold' linear structures for content analysis along the principal axis.

Main Results:

• Successfully computed centerlines for elongated and curved shapes in spatial transcriptomics data.
• Enabled the analysis of cell and transcript type proportions along the computed principal axis.
• Facilitated the study of spatial distribution of cells or transcripts perpendicular to the centerline.

Conclusions:

• The proposed method effectively addresses computational challenges in analyzing spatial transcriptomics data from specific tissue shapes.
• This technique enhances the ability to study cellular and molecular organization within linear tissue structures.
• Offers a new tool for detailed spatial analysis in transcriptomics research.