Cancer has been one of the leading causes of death in this century. We aim to improve the understanding of the biological processes underlying cancers which could possibly lead to identification of early biomarkers and personalized treatment. In this project, we focus on comparing spatial single cell RNA expressions between samples with various stages of cancers. We use neural networks to detect gene signatures that would drive the cell-cell communication within the malignant spots. Using these gene signatures, we infer their upstream regulators which could serve as candidates for biomarkers and drug targets. A human spatial as well as non-spatial scRNA expression data from various cancers from already published studies will be used in this project.