GITIII-scale is presented, a hierarchical, interpretable pan-cancer spatial transcriptomics foundation model for TME representation learning that investigates cell state-niche associations and their underlying ligand-receptor (LR) signaling pathways that recovered niche-associated state changes more accurately than existing spatial transcriptomics foundation models in cancer types unseen during training.
Abstract
In the tumor microenvironment, cell's state is influenced by cell-cell interactions (CCIs) with neighboring cells in its niches. Identifying dysregulated CCIs that are associated with pathogenic process pinpoints targets for drug discovery. Imaging-based spatial transcriptomics and single-cell RNA sequencing provide, respectively, single-cell spatial information and transcriptome-wide measurements needed to study CCIs, but neither modality provides both. Existing spatial transcriptomics foundation models also cannot effectively learn from spatially resolved single-cell data with full-transcriptome coverage, explicitly infer the CCI mechanisms driving cell state-niche associations, or interpretable enough to support direct biological interpretations. Here, we present GITIII-scale, a hierarchical, interpretable pan-cancer spatial transcriptomics foundation model for TME representation learning that investigates cell state-niche associations and their underlying ligand-receptor (LR) signaling pathways. GITIII-scale uses transformers to model interactions between pairs of cells at defined spatial distances, an interpretable single-layer graph transformer without a feed-forward network to decompose how each gene in a receiver cell is influenced by each neighboring sender cell, and a graph transformer to generate cellular-neighborhood embeddings. Trained on our assembled pan-cancer database of specimen-matched scRNA-seq and imaging-based spatial transcriptomics datasets, GITIII-scale generated TME embeddings that recovered niche-associated state changes more accurately than existing spatial transcriptomics foundation models in cancer types unseen during training. A case study of an unseen breast cancer dataset further demonstrated the model's interpretability by identifying potentially drug-targetable LR pathways associated with endothelial overgrowth and tumorigenesis.
A Framework for Learning Over REgulatory-Embedding Networks (FloREN), a supervised and interpretable sample representation method that enables improved sample stratification and biomarker discovery and supports downstream analyses that found specific immune network mechanisms in immune-mediated inflammatory diseases (IMIDs).
Iñigo Clemente‐Larramendi, S. Hillion, D. Cornec et al.· bioRxiv· 0 citations
Phenoverse is introduced, an interpretable deep learning framework that learns sample-level disease state representations through cell type-aware residual encoding, prototype learning, and Perceiver-based aggregation that demonstrates that trajectory-derived genes reveal cross-cohort molecular programs and show consistently higher reproducibility than traditional case-control comparisons.
Manoj M Wagle, Yongheng Wang, Soham Samanta et al.· bioRxiv· 0 citations
ACSCeND offers a robust, interpretable approach to profiling CSC dynamics and establishes CSC state as a clinically meaningful, pan-cancer biomarker for guiding stemness-informed therapies by integrating single-cell precision with bulk-level applicability.
Single-cell transcriptomics resolves CAR T-cell states, yet translating heterogeneous cellular signals into patient-level therapeutic response remains challenging. Existing studies primarily identify response-associated genes or cell populations through experimental and statistical analyses, but few predictive frameworks integrate gene-level structure with clinical outcomes. Here, we present gANCHOR, a T-cell foundation model built on a hierarchical hypergraph attention framework combining biologically informed representation learning with patient-level response prediction. By encoding gene-pathway relationships, gANCHOR learns pathway-aware cell embeddings that improve biological conservation and batch robustness. A cell-to-patient attention module then aggregates cellular information to infer therapeutic response. Across benchmark datasets, gANCHOR achieved the strongest overall performance in biological conservation and batch-correction assessments. In response prediction across 161 patients from five CAR T-cell studies, gANCHOR achieved an F1 score of 0.87, outperforming benchmarked single-cell foundation models. gANCHOR also identified reproducible response- and non-response-associated gene programs, providing interpretable biological insights into CAR T-cell efficacy and resistance.
Yawei Li, Deyu Fang, Chengsheng Mao et al.· bioRxiv· 0 citations
Spatial omics allows for comprehensive investigation of the tumor immune microenvironment (TIME). Stratifying patients by their TIMEs contributes with insights in tumor immune response and has the potential to guide treatment decision-making in the immuno-oncology setting. However, high-plex spatial omics approaches still suffer from high costs and limited direct clinical applicability. We address this issue by presenting a deep learning model, Image2Count, for deconvoluting molecular expression from low-plex immunofluorescence imaging. Explicitly, our model learns visual representations of cells in a contrastive manner, utilizing Graph Neural Networks to predict expressions from cell graphs, enabling the trained model to predict high-plex single-cell expression from just four marker images. We measure model performance using an ovarian cancer GeoMx dataset with “bulk” Region of Interest 72-plex protein counts, a Cyclic IF (t-CyCIF) 25-plex single-cell resolution colorectal cancer dataset, and a 960-plex RNA single-cell resolution CosMx non-small cell lung cancer dataset. Image2Count is able to predict distinct spatial expression patterns of subsets of tumor, immune and stromal cells, and displays a generally improved accuracy when considering neighborhoods of cells over single cells. Concordance of pathways enriched in true and predicted data indicates the ability to capture biologically relevant information. Our model paves the way for clinically implementable TIME stratification based on low-plex immunofluorescence images, and allows for standard single-cell analysis workflows to interpret multicellular expression data from regions of interest.
Markus Heidrich, Daniel Nilsson, A. M. Frank et al.· npj Precision Oncology· 0 citations
Spatial transcriptomics now profiles patient cohorts at single-cell resolution, enabling analysis of disease-associated cell organization in situ. However, discovering such spatial biomarkers remains challenging because relevant structures occur at unknown scales and cell-or niche-level annotations are rarely available. We present spHOT, a deep learning framework that localizes phenotype-associated spatial biomarkers from sample-level labels. spHOT combines spatial foundation model embeddings, a hierarchical domain tree for multi-resolution tissue representation, and a teacher-student multiple instance learning architecture that converts sample labels into cell-level biomarker scores. In controlled simulations and real-tissue benchmarks, spHOT outperformed existing spatial and single-cell methods in localizing ground-truth biomarkers. Across fibrotic, metabolic, and autoimmune disease datasets, spHOT recovered disease-relevant niches and tissue states reported by supervised analyses in the original studies. Cross-disease application of spHOT transferred biomarkers across chronic lung diseases without retraining. spHOT enables scalable, annotation-efficient spatial biomarker discovery in cohort-scale spatial transcriptomics.
H. Kim, Donghee Kim, Sangwook Jung et al.· bioRxiv· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
