Research map / Vision-language and multimodal / Human and embodied modeling
spora: A Unified Multimodal Dataset for Spatial Proteomics
Benedikt von Querfurth, Johann Wenckstern, Eeshaan Jain, Lukas Klein, Phil F Cheng, Yexiang Cheng, Cédric Vincent-Cuaz, Martina Haberecker, Pascal Frossard, Andreas Wicki, Olivier Michielin, Charlotte Bunne
Read the paperCheck your idea against 137,217 papers
AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Introduces spora, a unified multimodal dataset for spatial proteomics intended to support models that bridge heterogeneous protein panels, platforms, and tissue contexts.
Problem
Spatial proteomics cohorts differ in protein panels, acquisition platforms, and tissue contexts, making it hard to train models that yield a coherent picture of tissue biology.
Approach
A unified multimodal dataset for spatial proteomics that aims to bridge cohort-level heterogeneity so foundation models can be trained across panels, platforms, and tissues.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use VirTues instead of KRONOS for cell embeddings | linear probe · 10 cohorts cell phenotyping | macro F1: 0.355 -> 0.611 (cohort 3) | Improved |
| use VirTues over KRONOS for tissue-level prediction | ABMIL classifier · three diagnostic/therapeutic cohorts | performance | Mixed |
| use VirTues instead of Cellpose/InstanSeg/StarDist for zero-shot cell segmentation | segmentation · eight cohorts | F1 at IoU 0.5 | Improved |
| use pretrained H&E-to-multiplex virtual staining models on new cohorts | GigaTIME/HistoPlexer/ROSIE · four cohorts | Pearson: correlations 0.003 to 0.106 | Hurt |
| use spora[io] coverage tiler instead of grid tiling | tiling · example tissues | tissue-to-background ratio: ratio 5.43 to 6.18 | Improved |
| use VirTues vs astir and MAPS for cross-cohort annotation | annotation · eight cohorts | macro F1 | Mixed |
Closest papers by meaning
- HEST-1k: A Dataset For Spatial Transcriptomics and Histology Image Analysis (NeurIPS 2024 · similarity 0.90)
- SToFM: a Multi-scale Foundation Model for Spatial Transcriptomics (ICML 2025 · similarity 0.90)
- STimage-1K4M: A histopathology image-gene expression dataset for spatial transcriptomics (NeurIPS 2024 · similarity 0.88)
- Diffusion Generative Modeling for Spatially Resolved Gene Expression Inference from Histology Images (ICLR 2025 · similarity 0.88)
- M2OST: Many-to-one Regression for Predicting Spatial Transcriptomics from Digital Pathology Images (AAAI 2025 · similarity 0.87)
- Towards Scalable Context-Aware Single-Cell Spatial Transcriptomics Prediction from Histology Images (NeurIPS 2026 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
