Skip to content

RMGAT: a Relational Multi-Scale Graph Attention Network for spatial transcriptomics domain identification with systematic component ablation

Sep 2026 · BMC Bioinformatics · 0 citations
Single-cell and spatial transcriptomics

TL;DR

A systematic ablation study providing evidence-based design guidance for spatial transcriptomics GNNs is provided, with findings support using GAT attention and contrastive learning and investing in node feature quality over edge feature complexity.

Abstract

Mapping spatially coherent tissue domains from spatial transcriptomics data is a prerequisite for characterizing cell-type, developmental patterning, and disease-associated disruptions of tissue architecture. While graph neural network (GNN) methods have substantially improved domain identification over expression-only clustering, the effect of edge feature design (i.e. how spatial connections between tissue spots are represented) has never been systematically evaluated. All existing methods use a single scalar edge weight, leaving open the question of whether multi-dimensional edge features provide any benefit. We introduce RMGAT, a Relational Multi-Scale Graph Attention Network that applies multi-dimensional learned edge features to spatial transcriptomics, to the best of our knowledge for the first time. Each spatial edge is described by an 8-dimensional learned feature vector. We combine this with a three-component training objective (graph reconstruction, NT-Xent contrastive learning, and a self-expression decoder) and a V30 post-processing pipeline using consensus clustering, Hungarian label alignment, and spatial refinement. Evaluated on the 12-section DLPFC (human dorsolateral prefrontal cortex) benchmark across 60 independent runs (5 seeds × 12 DLPFC sections), RMGAT achieves a mean adjusted Rand index (ARI) of 0.3422 ± 0.0402 (95% confidence interval [0.332, 0.353]), comparable to SpaGCN (~ 0.36). A systematic four-component ablation study with formal statistical testing identifies GAT attention as the only statistically confirmed essential contributor (ΔARI = − 0.028 when replaced with a graph convolutional network, GCN; p  < 0.0001). Contrastive learning shows a directional benefit (ΔARI = − 0.017; p  = 0.024 vs. scalar-weight baseline). The 8-dimensional EdgeMLP and self-expression decoder are both neutral (ΔARI ≈ 0.004; both p  > 0.3), indicating neither contributes at PCA-50 input quality. The primary contribution of this work is a systematic ablation study providing evidence-based design guidance for spatial transcriptomics GNNs. At PCA-50 input quality, GAT attention is the statistically confirmed critical component; contrastive learning is beneficial ( p  = 0.024); multi-dimensional edge features and additional decoder objectives do not improve performance under this input regime. For practitioners, these findings support using GAT attention ( p  < 0.0001) and contrastive learning ( p  = 0.024, non-significant after Bonferroni correction), and investing in node feature quality over edge feature complexity.

Read PDF

Similar papers

#computer vision Review Open access Mar 2016

Determinants of overweight or obesity among ever-married adult women in Bangladesh

The findings of the study indicate that a large number of women in Bangladesh are suffering from being overweight or obese, and multiple factors are responsible for this including, older age,being from wealthy households, higher education, being from food-secured households, watching TV at least once a week, and being...

Haribondhu Sarma, N. Saquib, M. Hasan et al. · 75 citations · ⚡5
#artificial intelligence Open access Sep 2026

Generating Interesting Scientific Ideas using Knowledge Graphs and LLMs: Evaluations with 100 Research Group Leaders

The rapid growth of scientific literature makes it increasingly challenging for researchers to identify novel and impactful ideas, especially across disciplines. Modern artificial intelligence (AI) systems offer new opportunities for scientific ideation, but how compelling are AI-generated ideas, and how can their qual...

Xuemei Gu, Mario Krenn · 65 citations
#computer vision Conference Aug 2008

Scrum in a Multiproject Environment: An Ethnographically-Inspired Case Study on the Adoption Challenges

Agile methods continue to gain popularity. In particular, the Scrum method appears to be on the verge of becoming a de-facto standard in the industry, leading the so called Agile movement. While there are success stories and recommendations, there is little scientifically valid evidence of the challenges in the adoptio...

A. Marchenko, P. Abrahamsson · 59 citations · ⚡11
#data science Review Open access Apr 2022

Association between Mother’s Education and Infant and Young Child Feeding Practices in South Asia

Findings demonstrate a strong positive association between mother’s education and IYCF indicators and suggest strengthening national policies to educate women at least to the secondary level in SACs might be a cost-effective intervention for improving IY CF practices.

M. Tariqujjaman, M. Hasan, M. Mahfuz et al. · 45 citations · ⚡4
#computer vision Open access Sep 2012

Making the leap to a software platform strategy: Issues and challenges

A comprehensive taxonomy of the challenges faced when a medium-scale organization decided to adopt software platforms is provided, namely: business challenges, organizational challenges, technical challenges, and people challenges.

Yaser Ghanam, F. Maurer, P. Abrahamsson · 41 citations · ⚡3

Related blog posts

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.