REACTIONS IN DATABASES:

Reactions are stored in databases as TDTs. The normalizations "USMILES" and "ASMILES" apply as follows in Thor to handle reaction-rooted dataitems:

"Unique" reactions:

• No agents or stereo.
  
• Typically used as root of reaction TDT.

"Absolute" reaction:

• All components and stereo.
• Atom map values.
  
• Used either as root of subtrees or as data (preferred).

The result of these normalizations is that reactions are stored together on the same TDT page if they contain the same generic reactant and product molecules.

One of the potential difficulties here is database conventions for inclusion of non-key molecules. In the following case, these two esterifications will be stored on different TDT pages in the database:

$SMI<"CC(=O)O+OCC>>CC(=O)OCC">|
$SMI<"CC(=O)O+OCC>>CC(=O)OCC+O">|

The solution has two parts: first, develop and enforce conventions for registration. Second, lookup by both reactant and product in Thor.