SUICA: Learning Super-high Dimensional Sparse Implicit Neural Representations for Spatial Transcriptomics

2024-12-02

arXiv.org

AI Summary

Spatial transcriptomics (ST) data pose modeling challenges due to discreteness, ultra-high dimensionality, sparsity, and highly skewed gene expression distributions. To address these, we propose the first graph-augmented autoencoder-guided implicit neural representation (INR) framework, which achieves continuous, compact modeling via structure-aware embedding learning. We further introduce a novel regression-to-classification loss to mitigate fitting bias induced by the heavy-tailed expression distribution. Our method consistently outperforms existing INR variants and state-of-the-art methods across multiple ST platforms. It achieves significant improvements in numerical fidelity (PSNR/SSIM), gene–spatial statistical correlation (Moran’s I, Spearman correlation coefficient), and biological pathway consistency. Crucially, it enhances signal fidelity for key genes, thereby enabling more reliable downstream spatial decomposition and functional inference.

Technology Category

• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning
• Data Mining & Knowledge Management: Mining of Spatial, Temporal or Spatio-Temporal Data
• Machine Learning: Graph-based Machine Learning

Application Category

• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs
• Search and Retrieval-Augmented AI: Web query analysis, representation and understanding
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Spatial Transcriptomics (ST) is a method that captures gene expression profiles aligned with spatial coordinates. The discrete spatial distribution and the super-high dimensional sequencing results make ST data challenging to be modeled effectively. In this paper, we manage to model ST in a continuous and compact manner by the proposed tool, SUICA, empowered by the great approximation capability of Implicit Neural Representations (INRs) that can enhance both the spatial density and the gene expression. Concretely within the proposed SUICA, we incorporate a graph-augmented Autoencoder to effectively model the context information of the unstructured spots and provide informative embeddings that are structure-aware for spatial mapping. We also tackle the extremely skewed distribution in a regression-by-classification fashion and enforce classification-based loss functions for the optimization of SUICA. By extensive experiments of a wide range of common ST platforms under varying degradations, SUICA outperforms both conventional INR variants and SOTA methods regarding numerical fidelity, statistical correlation, and bio-conservation. The prediction by SUICA also showcases amplified gene signatures that enriches the bio-conservation of the raw data and benefits subsequent analysis. The code is available at https://github.com/Szym29/SUICA.

Problem

• Modeling super-high dimensional sparse ST data continuously
• Enhancing spatial density and gene expression using INRs
• Improving bio-conservation and analysis of spatial transcriptomics

Innovation

• Uses Implicit Neural Representations for spatial modeling
• Incorporates graph-augmented Autoencoder for context information
• Applies classification-based loss for skewed distribution