Breast cancer IMC dataset for InterSTELLAR training

<strong>graph_data.npy</strong>: preprocessed Breast cancer IMC dataset for InterSTELLAR training, saved as a<strong> .npy</strong> file. Read this file with python and get a list named <strong>all_graph</strong>. This list includes <strong>368 </strong>elements. The first <strong>366 </strong>elements are the information from each tissue, the last two elements are the <strong>mean </strong>and<strong> standard deviation</strong> of the cell feature data after log-transformation with shape <strong>30x1</strong>. Each sub-element <strong>all_graph[i]</strong> is alone a new list with three elements, including <strong>cell feature matrix all_graph[i][0]</strong>,<strong> cell locations all_graph[i][1],</strong> <strong>tissue labels all_graph[i][2] </strong>and <strong>cell phenotypes</strong> <strong>all_graph[i][3]</strong>. <strong>all_graph[i][0]</strong> is a <strong>Nx30</strong> matrix, corresponding to<strong> 30 cell markers</strong>; <strong>all_graph[i][1]</strong> is a<strong> Nx2</strong> matrix, corresponding to <strong>x </strong>and <strong>y </strong>locations of a single cell; <strong> all_graph[i][2]</strong> is a list with<strong> 5</strong> elements, corresponding to <strong>tissue phenotypes</strong> (0: healthy, 1: TNBC, 2: Non-TNBC cancers), <strong>tissue area</strong> (um²),<strong> tumor grade</strong>, <strong>overall survival time </strong>(month) and <strong>patient status</strong> (alive or death); <strong>all_graph[i][3] </strong>is a list with <strong>N </strong>elements corresponding to the <strong>cell phenotypes</strong>. <strong>cell_masks.npy</strong>: the cell segmentation maskes corresponding to the 366 tissues.