Generate multiple conformers
============================

In contrast to conformer searching, the goal of conformer generation is not simply to find a low energy conformation but to generate several different conformations. Such conformations may be required by another piece of software such as some protein-ligand docking and pharmacophore programs. They may also be useful if considering writing some sort of shape comparison method.

Open Babel has two distinct conformer generating codes:

1. Confab: A systematic conformer generator that generates all diverse
   low-energy conformers.

2. Genetic algorithm: This is a stochastic conformer generator that generates
   diverse conformers either on an energy or RMSD basis

Genetic algorithm
-----------------

A genetic algorithm is a general computational method to find a globally optimum solution to some multiparameter problem. It involves a population of conformers which after a series of generations, iteratively arrive at an optimal solution in terms of either RMSD diversity or energy.

Information about using this method is available at the command-line using: ``obabel -L conformer``. Although labelled as "Conformer Searching", if you choose the genetic algorithm method (which is the default) then you can save the conformers in the final generation using ``--writeconformers``. For example, the following line creates 30 conformers optimized for RMSD diversity::

  obabel startingConformer.mol -O ga_conformers.sdf --conformer --nconf 30
         --score rmsd --writeconformers

In this case ``--score rmsd`` was not strictly necessary as RMSD diversity was the default in any case.

.. _Confab:

Confab
------

Confab systematically generates all diverse low-energy conformers for molecules. To run Confab use the ``--confab`` operation, and to assess the results by calculating RMSDs to reference structures, use the **confabreport** output format.

.. rubric:: confab operator

*  ``obabel <inputfile> -O <outputfile> --confab [confab options]`` for typical usage
*  ``obabel -L confab`` for help text

The *inputfile* should contain one or more 3D structures (note that 2D structures will generate erroneous results). Generated conformers are written to the *outputfile*. All of the conformers for a particular molecule will have the same title as the original molecule.

--rcutoff <rmsd>
    RMSD cutoff (default 0.5 Angstrom)
--ecutoff <energy>
    Energy cutoff (default 50.0 kcal/mol)
--conf <#confs>
    Max number of conformers to test (default is 1 million)
--original
    Include the input conformation as the first conformer
--verbose
    Verbose - display information on torsions found

.. rubric:: confabreport format

*  ``obabel <inputfile> [-O <outputfile>] -o confabreport -xf <reference_file> [-xr <rmsd>]`` for typical usage
*  ``obabel -L confabreport`` for help text

Once a file containing conformers has been generated by Confab, the result can be compared to the original input structures or a set of reference structures using this output format. Conformers are matched with reference structures using the molecule title. For every conformer, there should be a reference structure (but not necessarily *vice versa*).

-f <filename>
     File containing reference structures
-r <rmsd>
     RMSD cutoff (default 0.5 Angstrom)

     The number of structures with conformers within this RMSD cutoff
     of the reference will be reported.

.. rubric:: Example

The example file, `bostrom.sdf`_, contains 36 molecules which have from 1 to 11 rotatable bonds (see *Bostrom, Greenwood, Gottfries, J Mol Graph Model, 2003, 21, 449*).

We can generate and test up to 100K conformers using Confab as follows::

  > obabel bostrom.sdf -O confs.sdf --confab --conf 100000

  **Starting Confab 1.1.0
  **To support, cite Journal of Cheminformatics, 2011, 3, 8.
  ..Input format = sdf
  ..Output format = sdf
  ..RMSD cutoff = 0.5
  ..Energy cutoff = 50
  ..Conformer cutoff = 1000000
  ..Write input conformation? False
  ..Verbose? False

  **Molecule 1
  ..title = 1a28_STR_1_A_1__C__
  ..number of rotatable bonds = 1
  ..tot conformations = 12
  ..tot confs tested = 12
  ..below energy threshold = 10
  ..generated 3 conformers

  ... etc, etc

  0 molecules converted

To check how many of the generated conformers are within 1.0 A RMSD of the original structures, we can use the confabreport format as follows::

  > obabel confs.sdf -oconfabreport -xf bostrom.sdf -xr 1.0

  **Generating Confab Report
  ..Reference file = bostrom.sdf
  ..Conformer file = confs.sdf

  ..Molecule 1
  ..title = 1a28_STR_1_A_1__C__
  ..number of confs = 3
  ..minimum rmsd = 0.0644801
  ..confs less than cutoffs: 0.2 0.5 1 1.5 2 3 4 100
  ..1 1 3 3 3 3 3 3
  ..cutoff (1) passed =  Yes

  ... etc, etc

  **Summary
  ..number of molecules = 36
  ..less than cutoff(1) = 35
  52271 molecules converted

.. _bostrom.sdf: ../_static/bostrom.sdf