Similarity searching

Introduction

In order to be able to set up searches we will need to import the ccdc.search module. Let us also import the ccdc.io module to allow us to read in and write out molecules.

>>> import ccdc.search
>>> import ccdc.io

As a preamble let us also set up a variable for a temporary directory and a file path to a testosterone molecule.

>>> import tempfile
>>> tempdir = tempfile.mkdtemp()
>>> filepath = 'testosterone.mol2'

To get access to the molecule in the testosterone mol2 file we make use of a ccdc.io.MoleculeReader.

>>> reader = ccdc.io.MoleculeReader(filepath)
>>> testosterone = reader[0]

See also

API documentation of the search module

Similarity search

To run a similarity search one must first create a ccdc.search.SimilaritySearch whose initialiser takes a ccdc.molecule.Molecule and a similarity threshold between 0.0 and 1.0. By default the similarity threshold is set to 0.7.

>>> similarity_query = ccdc.search.SimilaritySearch(testosterone)

The similarity search can then be run by making use of the search() function.

>>> sim_hits = similarity_query.search()
>>> print(len(sim_hits))
796

To reduce the number of hits we can increase the similarity threshold.

>>> similarity_query.threshold = 0.9
>>> sim_hits = similarity_query.search()
>>> print(len(sim_hits))
83

An alternative approach to reducing the number of hits is to constrain the number of hits to return.

>>> sim_hits = similarity_query.search(max_hit_structures=10)
>>> print(len(sim_hits))
10

Let us find out what these structures are and what their similarity to the query is.

>>> for hit in sim_hits:
...     print('%9s: %.2f' % (hit.identifier, hit.similarity))
   BAWMAN: 1.00
   BEJVAN: 1.00
   BOKVUS: 1.00
   CERVAX: 1.00
   DIGRIV: 1.00
   EFEJAD: 1.00
   EPITES: 1.00
   GIXVIW: 1.00
   GIXVOC: 1.00
   HANSTO: 1.00

Similarity searches allow all forms of search filter given by ccdc.search.Search.Settings. See Search filters for examples of use.

Similarity queries allow all the forms of search that a ccdc.search.SubstructureSearch does.

See also

For more information please see Setting up and running a substructure search

For example we can use an instance of a ccdc.molecule.Molecule directly.

>>> file_path = 'ABEBUF.mol2'

To get access to the first molecule in the ABEBUF.mol2 file we make use of the ccdc.io.MoleculeReader.

>>> h = similarity_query.search_molecule(ccdc.io.MoleculeReader(file_path)[0])
>>> print('Identifier: %s, similarity: %.3f' % (h.identifier, h.similarity))
Identifier: ABEBUF, similarity: 0.106