Grant: Project Grant

Synovial sarcoma (SS) is an aggressive cancer that typically strikes adolescents and young adults with long-term poor outcomes as current drugs are ineffective. It arises from a single genetic error - a fusion gene called SS18::SSX, yet it gives rise to a wide range of tumor forms, often evading the body's immune defenses. At the same time, because of its relatively simple molecular makeup, SS is an ideal model for developing AI-driven methods that can later be applied to more common and complex cancers. Our project capitalizes on exclusive genomic profiling and editing resources of SS - including novel data from a large set of human primary tumors and a deeply profiled genetically engineered mouse model (GEMM) - to map how SS cells regulate gene expression, remodel their DNA, and interact at different times and locations. We integrate these extensive datasets into a predictive and explainable AI model that recognizes which normal developmental pathways have been hijacked by the tumor, and how this hijacking leads to immune evasion. By continuously learning from new experiments, our AI model will highlight the most promising targets for therapy - genes or pathways that, if properly targeted, may restore the immune system's capability to attack the SS tumor. We will validate these targets using genetic perturbation and spatial mapping in mouse models and patient-derived SS systems. Our pilot gene-perturbation study links selected genetic changes to measurable single-cell responses in SS, providing a causal cross-check and validation for the AI's top nominations. Consistent with these nominations, perturbing a leading immune-evasion circuit improves survival in our mouse model. Ultimately, we aim to create an iterative AI-experiment loop that refines mechanisms of SS tumourigenesis, leading to therapeutic strategies for the treatment of SS and improving the outlook for young people facing this devastating disease.