2.5. Chemical Space Visualization

Morgan fingerprints with radius 2 (Morgan2, 1024-bits) were generated for each compound and fragment data set. To generate a visual representation of the chemical space, we used the recently developed algorithm TMAP (Tree MAP). This method allows the visual representation of many molecules that are difficult to visualize using other standard methods such as principal component analysis. Basically, TMAP allows the visualization of large data sets (such as the ones studied in this work—Table 1) through the distance between the clusters and the cluster’s detailed structure through branches and sub-branches [26,27]. Fingerprints for each data set (input data) were indexed in a local sensitive hashing (LSH) forest data structure, enabling c-approximate k-nearest neighbor (k-NN). Fingerprints were encoded using the MinHash algorithm. An undirected weighted c-approximate k-nearest neighbor graph (c-k-NNG) is constructed from the data points indexed in the LSH forest. This graph takes two arguments, k, the number of nearest-neighbors, and kc, the factor used by the augmented query algorithm. In this work, we used k = 50 and kc =10. Further details of the TMAP approach are published elsewhere [28].