Measuring similarity between molecules is an important part of virtual screening (VS) experiments deployed during the early stages of drug discovery. Most widely used methods for evaluating the similarity of molecules use molecular fingerprints to encode structural information. While similarity methods using fingerprint encodings are efficient, they do not consider all the relevant aspects of molecular structure. In this paper, we describe a quantum-inspired graph-based molecular similarity (GMS) method for ligand-based VS. The GMS method is formulated as a quadratic unconstrained binary optimization problem that can be solved using a quantum annealer, providing the opportunity to take advantage of this nascent and potentially groundbreaking technology. In this study, we consider various features relevant to ligand-based VS, 1 arXiv:1902.00352v1 [q-bio.QM] 28 Jan 2019 such as pharmacophore features and three-dimensional atomic coordinates, and include them in the GMS method. We evaluate this approach on various datasets from the DUD_LIB_VS_1.0 library. Our results show that using three-dimensional atomic coordinates as features for comparison yields higher early enrichment values. In addition, we evaluate the performance of the GMS method against conventional fingerprint approaches. The results demonstrate that the GMS method outperforms fingerprint methods for most of the datasets, presenting a new alternative in ligand-based VS with the potential for future enhancement. a) Step 1: Modelling molecules as graphs G 1 G 2 Conflict graph (G 1 ,G 2 ) Solution to the co-k-plex problem Map solution back to graphs 4 Molecule 1 Molecule 2 b) Step 2: Solving the co-k-plex problem c) Step 3: Similarity measure Figure 1: Illustration of the GMS method: a) Two molecules modelled as graphs. b) A conflict graph is built, the co-k-plex problem of the conflict graph is solved, and the solution is mapped back to the molecules. c) The similarity score is calculated. molecules in terms of the core scaffold fragment and its substructures, whereas physical chemists may focus on physicochemical properties, such as excluded volume and electrostatic properties. The similarity method introduced by Hernandez et al. 13 compares molecules according to chemical descriptors. In this work, we have modified the criteria for considering two molecules as similar by including weighted pharmacophore features. Additionally, the new similarity criteria allow the inclusion of partial matches. In this section, we describe how features and their relevance are incorporated in the GMS method.The GMS method consists of three steps: 1) modelling molecules as graphs; 2) solving the co-k-plex problem; and 3) calculating a coefficient that measures the similarity between the two input molecules. We describe these three steps in an earlier paper. 13 In the sections that follow, we present an overview of these processes with a special focus on the variations implemented in this study. A general scheme of this method is shown in Fig. 1.