MerQuaCo: a computational tool for quality control in image-based spatial transcriptomics

Area of Science:

• Spatial transcriptomics
• Genomics
• Bioinformatics

Background:

• Image-based spatial transcriptomics enables cell population identification and gene expression analysis in intact tissues.
• Analysis of large, high-dimension spatial datasets is facilitated by open-source software.
• Current methods for assessing spatial transcriptomics data quality are manual, subjective, and labor-intensive.

Purpose of the Study:

• To characterize common imperfections in spatial transcriptomics datasets.
• To assess the impact of these imperfections on downstream analyses, such as cell type labeling.
• To develop an automated tool for quantifying data quality in spatial transcriptomics.

Main Methods:

• Analysis of 641 fresh-frozen adult mouse brain sections from the Vizgen MERSCOPE platform.
• Identification and quantification of common data imperfections including tissue loss, detachment, dropped images, and variable detection probability.
• Development of MerQuaCo, an open-source computational tool for automated detection and quantification of spatial transcriptomics data imperfections.

Main Results:

• Common imperfections identified include local tissue loss, detachment from coverslip, dropped images, spatial variation in detection probability, and inter-experiment differences.
• These imperfections can significantly impact the accuracy of cell type labels derived from spatial transcriptomics data.
• MerQuaCo successfully detects and quantifies these imperfections without user input, enabling objective data quality assessment.

Conclusions:

• Spatial transcriptomics data quality assessment is crucial for accurate biological interpretation.
• MerQuaCo provides a rigorous and objective method for evaluating spatial transcriptomics data quality.
• This tool facilitates the selection of high-quality sections for further analysis, improving the reliability of spatial transcriptomics studies.