Seeing or believing in hyperplexed spatial proteomics via antibodies: New and old biases for an image-based...

Area of Science:

• Proteomics
• Cell Biology
• Immunohistochemistry

Background:

• Hyperplexed in-situ targeted proteomics (>15 markers) is revolutionizing cell and tissue classification.
• Unlike other high-dimensional single-cell assays, it relies heavily on human image evaluation for segmentation, thresholding, validation, and rendering.
• Established methods may introduce biases due to the limitations of human visual perception in this context.

Purpose of the Study:

• To re-evaluate all procedural steps in hyperplexed proteomics.
• To assess the limitations of human visual discrimination in image analysis.
• To introduce and evaluate BRAQUE, a tool for extracting granular information from low signal-to-noise markers.

Main Methods:

• Re-evaluation of human image analysis steps in hyperplexed proteomics.
• Assessment of human eye's gray level and luminance discrimination capabilities.
• Application of the BRAQUE tool to a public human lymph node dataset.
• Analysis of antibody staining results for specificity.

Main Results:

• The human eye has limited capacity to discriminate gray levels (<64/256) and luminance in histology images.
• BRAQUE successfully extracts marker-agnostic information from low signal-to-noise ratio markers not visible through traditional rendering.
• Analysis revealed unpredicted staining results from validated antibodies, questioning current definitions of antibody specificity.

Conclusions:

• Human visual assessment in hyperplexed proteomics is limited and potentially biased.
• BRAQUE offers a method to overcome visual limitations and extract deeper biological insights.
• There is a critical need to redefine antibody specificity standards for hyperplexed immunostaining.