Nextflow Pipeline for Visium and H&E Data from Patient-Derived Xenograft Samples

Area of Science:

• Computational Biology
• Genomics
• Bioinformatics

Background:

• Spatial transcriptomics and histology imaging are crucial for understanding tissue architecture and gene expression.
• Analyzing patient-derived xenograft (PDX) models requires robust methods to differentiate host and graft data.
• Integrating multi-modal data from spatial transcriptomics and hematoxylin and eosin (H&E) imaging presents analytical challenges.

Approach:

• Developed a Nextflow DSL2-based pipeline, Spatial Transcriptomics Quantification (STQ), for simultaneous processing of Visium spatial transcriptomics and matched H&E whole slide images.
• Implemented read deconvolution for host (mouse) and graft (human) species using Xenome and gene expression quantification with Space Ranger.
• Integrated H&E image processing to generate morphometric and deep learning features per Visium spot, enabling multi-modal comparisons.

Key Points:

• The STQ pipeline enables accurate quantification of spatial gene expression and imaging features from PDX samples.
• Simultaneous analysis of transcriptomic and H&E data reveals complementary biological insights.
• The pipeline supports flexible analysis workflows, including single or dual reference genomes and standalone image analysis.

Conclusions:

• The developed pipeline provides a scalable, portable, and user-friendly framework for integrated spatial transcriptomics and H&E image analysis.
• Demonstrated the pipeline's utility on melanoma PDX samples, showing concordant clustering patterns between spatial gene expression and H&E imaging features.
• This approach facilitates multi-modal comparisons, enhancing the understanding of tumor microenvironments and PDX models.