Query molecules and hit compounds grouped in structural families. The 6 starting molecules known to interact with FV and/or FVIII (the queries) from previous studies can be grouped into 4 families (named here as clusters 1, 2, 3, and 4). We used these starting structures to search new molecules that could be structurally similar to the queries or relatively different (ie, both types of molecules are relevant). If we group the 9 active molecules found in our present study, we notice that 3 new compounds are related to cluster 1 (this one contains 2 members) (A), 2 new molecules are close to 2 query molecules grouped here under the name of cluster 3, whereas 1 new molecule is similar to 1 query compound (named here as cluster 4) (C). We found 3 other new molecules that are structurally unrelated to our starting queries. These 3 compounds can be grouped structurally into 2 families, one containing 2 members (6656941 and 6844287) and the other, only 1 member (5941257) (D). Cluster 2 did not result in the identification of new molecules (B). The reasoning behind such organization of the molecular structures is that it helps to rationalize the activity and is important for future optimization.