Fig. 2

Visualization of MCS-based rebalancing, on the reaction of ethyl acetate to ethanol. The first step of rebalancing is identifying the best alignment of product and reactant graphs (dotted black lines) corresponding to a maximum common subgraph. Subtracting the common structure leaves the raw missing compound, here the acetaldehyde shown in red on the left side. The dashed red line indicates the broken connection between the missing compound and the maximum common subgraph. Adding the missing compound may not be sufficient to obtain a chemically reasonable result. Thus a set of expand and merge rules are applied to fix the raw structure. In this example acetaldehyde is implausible. The lack of a second reactant triggers an expand rule that appends necessary oxygen, which is then attached to the new compound by a merge rule. In this case the merge rule creates a single bond between the two fragements, depicted by the dashed green line. This results in acetic acid as product. A valid solution is carbon balanced but not necessarily balanced after this step. The Rule-based method is therefore used to impute the missing water on the reactant side