Alternative normalization and analysis pipeline to address systematic bias in NanoString GeoMx Digital Spatial...

Area of Science:

• Genomics
• Molecular Biology
• Bioinformatics

Background:

• Spatial transcriptomics offers RNA expression data with tissue context.
• Quality assessment of end-user generated spatial transcriptomics data is often lacking.
• The NanoString GeoMx Digital Spatial Profiling (DSP) platform is a key technology in this field.

Purpose of the Study:

• To evaluate the quality of NanoString GeoMx DSP data and standard processing pipelines.
• To identify and address technical biases in spatial transcriptomics data.
• To propose improved normalization and analysis strategies for NanoString GeoMx DSP data.

Main Methods:

• Analysis of 72 regions of interest (ROIs) from 12 glioma samples.
• Replicate experiments for validation and evaluation of five external datasets.
• Comparison with bulk RNA sequencing and application of weighted gene co-expression network analysis.

Main Results:

• Consistent signal intensity variations across samples and experimental conditions were observed, leading to biased analyses.
• Quantile normalization effectively addressed technical differences in data distributions.
• NanoString DSP data exhibited a limited dynamic range compared to bulk RNA sequencing, potentially underestimating condition-specific differences.

Conclusions:

• NanoString GeoMx DSP data necessitate specialized normalization methods beyond standard pipelines.
• Alternative analysis strategies, including weighted gene co-expression network analysis, can extract meaningful biological insights.
• Robust quality assessment and tailored data processing are essential for reliable spatial transcriptomics research in oncology.