{"id":"arxiv:2607.09166","kind":"article","title":"COAST: Context-Aware Differential Learning for Gene Expression Prediction in Spatial Transcriptomics","authors":["Keunho Byeon","Sunhong Park","Jeewoo Lim","Jin Tae Kwak"],"author_count":4,"year":2026,"published":"2026-07-10","venue":"Lecture notes in computer science","cited_by":1,"doi":"10.1007/978-3-032-38224-5_6","arxiv_id":"2607.09166","nct_id":null,"source_url":"https://arxiv.org/abs/2607.09166","archive_url":"https://space-kitty.com/papers/arxiv:2607.09166","updated":"2026-07-10","pmid":null,"pmcid":null,"openalex_id":"W7168125216","pdf_url":"https://arxiv.org/pdf/2607.09166","abstract":"Spatial transcriptomics enables profiling of spatial gene expression but is limited by high cost and low throughput, motivating prediction from H&E histopathology images. Existing context-aware methods mainly supervise absolute expression, while relative expression relationships between spots are rarely used explicitly. We propose COAST, a context-aware differential learning framework for spatial gene expression prediction. COAST conditions the local and global context features with type-specific modulation and aggregates the target and context spot tokens using a Transformer encoder to capture both fine-grained local patterns and slide-level structure. It is trained with a joint objective that combines absolute expression regression with signed differential regression between the target and context spots. Experiments on multiple spatial transcriptomics datasets show consistent improvements in correlation- and distribution-based metrics, demonstrating the effectiveness of context-aware differential learning for histology-based spatial gene expression prediction.","abstract_license":"CC0-1.0 (arXiv descriptive metadata)","license":null,"is_open_access":false,"retracted":false,"peer_reviewed":null,"categories":["cs.LG"],"details":null,"provenance":"arxiv","attribution":"Abstract: arXiv descriptive metadata (CC0). Thank you to arXiv for use of its open access interoperability. This service was not reviewed or approved by, nor does it necessarily express or reflect the policies or opinions of, arXiv.","topics":[{"slug":"single-cell-spatial","name":"Single-Cell & Spatial Omics","domain_slug":"biotech"}],"cited_in_dossiers":[]}