A simultaneously transmitting and reflecting RIS (STAR-RIS) empowered integrated sensing and communications (ISAC) framework is proposed, where the STAR-RIS establishes an additional link to compensate for the insufficient LoS link. To alleviate the conflicts between the limited wireless resources and the multifunctionality requirements, a cluster-based NOMA transmission scheme is adopted, where the communication functionality is employed by the joint communication and sensing (C&S) beam in a NOMA approach. A minimum beampattern gain maximization problem is formulated to jointly optimize the power allocation, active and passive beamformer (BF) design. We propose a block coordinate descent (BCD) based iterative algorithm, which splits the optimization variables into two blocks. For the joint power allocation and active BF block, the semidefinite relaxation and successive convex approximation are employed. For the passive BF block, the penalty-based method is invoked to deal with the non-convex constraints. Simulation results verified that our proposed algorithm achieves higher beampattern gain at the intended targets than the other baselines accompanying the least mismatch error.