Fig. 2

Main sub-layer composition: A layer starts with its depth, which is the distance of its atoms from the reaction center (which has depth = 0). The next three characters characterize the atom. The first character indicates the highest bond status among all bonds connected to this atom. (Additional file 1: Table S7. For instance, the Si atom (90E) is connected by 4 single bonds: 2 are not part of the reaction centre and are encoded by 0; 1 bond is broken, encoded by 7; and another one is made, encoded by 9, which is the highest bond status. The two other characters encode the atom symbol (Additional file 1: Table S8). E.g., 0E stands for Si. The connection table is contained between the brackets. Each bond in the connection table is encoded by 4 characters. E.g., the connection table (11GU11GS) encodes 2 bonds. The first two characters indicate the bond order in reactants for the first one and in products for the second one, respectively (Additional file 1: Table S6). 01GG means a bond is made with the atom at index GG. The last two characters represent the index of the other atom to connect to (see Additional file 1: Table S1 for the reaction center and the remaining group and Additional file 1: Table S2 for the leaving group). The square brackets store the atom stoichiometry, i.e. the number of times a same atom is in the products (Example in Additional file 1: Figure S1)