In a proof of principle study, we have applied an automated fluorescence toponome imaging system (TIS) to examine whether TIS can find protein network structures, distinguishing cancerous from normal colon tissue present in a surgical sample from the same patient. By using a three symbol code and a power of combinatorial molecular discrimination (PCMD) of 2(21) per subcellular data point in one single tissue section, we demonstrate an in situ protein network structure, visualized as a mosaic of 6813 protein clusters (combinatorial molecular phenotype or CMPs), in the cancerous part of the colon. By contrast, in the histologically normal colon, TIS identifies nearly 5 times the number of protein clusters as compared to the cancerous part (32 009). By subcellular visualization procedures, we found that many cell surface membrane molecules were closely associated with the cell cytoskeleton as unique CMPs in the normal part of the colon, while the same molecules were disassembled in the cancerous part, suggesting the presence of dysfunctional cytoskeleton-membrane complexes. As expected, glandular and stromal cell signatures were found, but interestingly also found were potentially TIS signatures identifying a very restricted subset of cells expressing several putative stem cell markers, all restricted to the cancerous tissue. The detection of these signatures is based on the extreme searching depth, high degree of dimensionality, and subcellular resolution capacity of TIS. These findings provide the technological rationale for the feasibility of a complete colon cancer toponome to be established by massive parallel high throughput/high content TIS mapping.