Python interface

This page gives a quick overview of how to use the Python classes and methods to compute distances between routes.

Input format

The routes or reaction tree are input as dictionaries that can be loaded from e.g. a JSON file. The structure of the dictionary follows the output format used by AiZynthFinder, but only a small number of fields are necessary. The structure is checked by the code, and if the validation fails, an exception will be raised.

The structure definition and validation code can be found in the route_distance.validation module.

The input structure should be a the type MoleculeNode which is a dictionary with the following fields

• smiles - the SMILES string of the molecule represented by the node

• type - should be the string “mol”

• children - an optional list, containing at most one dictionary of type ReactionNode

The ReactionNode type is a dictionary with the following fields

• type - should be the string “reaction”

• children - a list of dictionaries of type MoleculeNode

It is easy to realize that this is a recursive definition. All extra fields in the dictionaries are ignored. And example dictionary is shown below

{
    "smiles": "CCCCOc1ccc(CC(=O)N(C)O)cc1",
    "type": "mol",
    "children": [
        {
            "type": "reaction",
            "children": [
                {
                    "smiles": "CCCCOc1ccc(CC(=O)Cl)cc1",
                    "type": "mol"
                },
                {
                    "smiles": "CNO",
                    "type": "mol"
                }
            ]
        }
    ]
}

Calculating TED

To compute the distance between two routes, we will first load them from a JSON file

import json
with open("my_routes.json", "r") as fileobj:
    routes = json.load(fileobj)

Then we will create two wrapper objects that will take care of the calculations

from route_distances.ted.reactiontree import ReactionTreeWrapper
wrapper1 = ReactionTreeWrapper(routes[0], content="both")
wrapper2 = ReactionTreeWrapper(routes[1], content="both")

The argument content can be molecules, reactions or both and specify what nodes are included in the calculation.

To calculate the distance between these two routes, we simple do

wrapper1.distance_to(wrapper2)

If we have many routes, a distance matrix can be calculate using

from route_distances.ted.distances import distance_matrix
distance_matrix(routes, content="both")

The docstrings of all modules, classes and methods can be consulted here