MeowCat demo datasets — Visium / Xenium H&E examples (1–4, 6)

Demo input data for MeowCat (Multi-resolution Omics-informed Whole-slide Cell Annotation Tool), a deep-learning framework that predicts cell-type distributions across whole-slide H&E images using spatially-registered transcriptomics as supervision. This record bundles six example workflows covering single- and multi-patient training plus inference on new images: 01_visium_only (single Visium slide, RCTD soft labels), 02_xenium_only (single Xenium slide), 03_visium_xenium_single (one Visium + one Xenium slide), 04_multi_visium (4 LUAD Visium slide), 04_multi_xenium (4 LUAD Xenium slides), and 06_predict_new_sample (held-out H&E + a pre-trained checkpoint for inference-only demos). Each tarball contains only the raw inputs the MeowCat pipeline needs: raw H&E images, 10x Space Ranger / Xenium output, slimmed cell-bin alignments, pre-computed RCTD outputs and per-sample pixel-size-raw.txt . Larger multi-patient training (4 Visium + 4 Xenium) is hosted in a companion Zenodo record. Run with the MeowCat code at https://github.com/liranmao/MeowCat.