Genetics

JASPR: Joint Spatial Representation learning of histology and spatial genomics for improved virtual genomic screening and clinical prognostication

Marija Pizurica, Eric Zimmermann, Neil Tenenholtz, James Hall, Olivier Gevaert, Ava P. Amini, Lorin Crawford, Kristen A. Severson

Featured July 7, 2026

Original Paper

This analysis was generated by SciGrove. Upload your own PDFs or enter a DOI — and get the same AI breakdown on any paper.

Get started

AI-generated analysis — This is SciGrove's AI interpretation of the paper, not peer-reviewed content. Always refer to the original paper.

Simply

A new AI method called JASPR learns to connect detailed tissue pictures with gene activity maps by filling in missing information, helping doctors predict gene patterns and disease outcomes better.

In depth

The paper introduces JASPR, a self-supervised deep learning framework that integrates histology images (HE) and spatial transcriptomics (ST) data. It achieves this by employing a cross-modal reconstruction objective within a multiway transformer architecture, which incorporates spatial context across modalities. JASPR uses shared modules for universal spatial properties and modality-specific experts for unique features, leading to improved virtual genomic screening and prognostic value.

Key Takeaways

1
JASPR effectively integrates histology (HE) and spatial transcriptomics (ST) data by explicitly modeling spatial context within and across modalities.
2
The framework utilizes a masked autoencoder (MAE) objective within a multiway transformer, featuring shared attention mechanisms and modality-specific feed-forward networks.
3
JASPR significantly improves virtual genomic screening (predicting 9,248 genes from HE) and provides enhanced prognostic value for breast cancer outcomes, outperforming non-spatial and contrastive baselines.

Conceptual Flow

HIGH LEVEL

Methodology: How JASPR Integrates Data

The system takes pictures and gene maps, hides parts, and learns to guess the hidden parts by looking at both, making a smart combined understanding.

Tissue Picture

Gene Activity Map

Hide Parts & Guess

Combined Understanding

Results: Improved Predictions

By learning this way, the system got much better at predicting gene activity from pictures and forecasting patient outcomes.

Old Prediction Method

JASPR's Prediction Method

Compare Accuracy

Much Better Predictions