Strand AI releases virtual spatial proteomics maps for TCGA colon cancer: 442 H&E slides, seven predicted markers, browsable online
Synopsis
Strand AI ran its virtual staining model Lattice over the TCGA colon adenocarcinoma cohort to produce predicted spatial protein maps for 442 diagnostic H&E slides and released them publicly. The resource covers seven protein markers, CD3e, CD8, CD20, CD68, Ki67, CD45 and PanCK, plus DAPI, and lets users compare original tissue with predicted channels over the same field of view without signing in, or access it through Hugging Face or a Python SDK. The release states that these predictions carry a prognostic signal in this cohort, with details to follow.
Interpretation
Slides from a public colon adenocarcinoma cohort now come with companion predicted spatial protein maps, so a researcher can view the original H&E stain beside predicted protein channels over the same field of view. Previously the cohort offered the diagnostic H&E slides themselves, and protein-level spatial information had to be measured separately. The maps cover 442 diagnostic H&E slides from the TCGA-COAD cohort and were generated by the virtual staining model Lattice.
The maps give the spatial distribution of protein markers as per-pixel predictions rather than a single summary value for a whole slide. The virtual staining model Lattice maps an H&E image to predicted signal for specified protein markers, and this run selected seven markers plus DAPI. The included markers are CD3e, CD8, CD20, CD68, Ki67, CD45, PanCK and DAPI, and the browsing interface supports panning, zooming and channel toggling.
The resource is released in a form that can be retrieved reproducibly, with metadata and downloadable examples. Predictions and metadata are under CC BY-NC 4.0, which requires attribution and excludes commercial use; commercial use requires contacting Strand AI. The source H&E slides remain available from the NCI Genomic Data Commons under GDC data policies, and the resource is labeled research use only, not for clinical or diagnostic procedures.
Perspective
The demonstrated scope is 442 diagnostic H&E slides from the TCGA-COAD cohort with predicted maps for seven protein markers plus DAPI, intended for research use; the release explicitly excludes clinical or diagnostic procedures.
The release states that the predictions carry a prognostic signal in this cohort but says details will follow, and no specific analysis or numbers are given yet.