In the field of drug discovery, it is necessary to speed up the exploration and analysis of very large datasets of molecules. A possible approach is the clustering of molecules with similar structures to limit the exploration of the chemical-space. This computationally-intensive operation can be mapped to the quantum domain offering inherent parallelism and providing possible higher-quality solutions. We propose in this article a novel quantum-based clustering pipeline based on the use of quantum version of Generative Adversarial Network (QuGAN) to perform the clustering of open-access molecular data. Our contribution capitalizes on a previously-developped hybrid quantum data clustering algorithm that incorporates a molecule-specific quantum generative model for small molecular graphs (e.g.: QuMolGAN [11]). We present preliminary results based on noiseless simulations of a quantum processing unit with 4 qubits. It shows gains in clustering quality after training our quantum generative pipeline on graph-based data from a quantum mechanical dataset (QM9) of small organic molecules. This motivates further investigation, in particular the fine-tuning hyper-parameter optimization of the generative model, to achieve more distinct cluster separation and increased intra-cluster homogeneity.