Distributed multiple-input multiple-output (MIMO) is a type of the centralized radio access network (C-RAN) to achieve a high link capacity uniformly over the service area. A large-scale distributed MIMO requires prohibitively high computational complexity. Clustering of user equipments (UEs) can reduce the computational complexity, however, this produces the inter-cluster interference (ICI). In this paper, we propose the UE clustering and antenna selection beyond the cluster boundary assuming multi-user minimum mean square error filtering combined with singular value decomposition (MMSE-SVD). We exploit the multiplexing-diversity tradeoff in MMSE-SVD to suppress effectively the ICI while keeping the multi-user multiplexing capability. It is shown that the proposed UE clustering and antenna selection can achieve 6 times higher sum capacity and 1.6 times higher UE capacity, compared to the without clustering.