Critical cyber-physical systems have an increasingly significant role in the world, and ensuring their safety is a high priority objective. State of the art approaches and engineering tools can support the development process from very early stages, with high-level system modeling, analysis capabilities, and exploration of alternatives. However, these approaches are limited when it comes evaluation of complex extra-functional characteristics over designs with uncertainties, typical to early system designs. In my thesis project, I intend to introduce probabilistic graph queries for high level, scalable probabilistic analysis, fir for analysing system models with design uncertainty and applicable in design space exploration. The approach will be evaluated on external case studies, focusing on key performance metrics related to applicability in the target context, such as runtime, precision and formal guarantees.