Intro to Open Babel

Open Babel
Access and interconvert chemical
information

Noel M. O’Boyle
Open Babel development team and NextMove Software, Cambridge, UK

Nov 2012
Secret UK Location

Image credit: AJ Cann (AJC1 on Flickr)

Image credit: Jon Osborne (jonno101101 on Flickr)

• Volunteer effort, an open source success story
– Originally a fork from OpenEye’s OELib in 2001
– Lead is Geoff Hutchison (Uni of Pittsburgh)
– 4 or 5 active developers – I got involved in late 2005

• http://openbabel.org
• Associated paper: (Open Access)
– Open Babel: An open chemical toolbox, J. Cheminf., 2011, 3,
33.

5

Does anyone else use Open Babel?

• 40K downloads (from SF) in last 12 months
– 1.4K downloads of Windows Python bindings
• Paper #1 most accessed in last year
– Cited 60 times in 1 year
• In short, very widely-used

Features
• Multiple chemical file formats (+ options) and utility
formats
• 2D coordinate generation and depiction (PNG and
SVG)
• 3D coordinate generation, forcefield minimisation,
conformer generation
• Binary fingerprints (path-based, substructure-
based) and associated “fast search” database
• Bond perception, aromaticity detection and atom-
typing
• Canonical labelling, automorphisms, alignment
• Plugin architecture
• Several command-line applications, but also a
software library
• Written in C++ but bindings in several languages

obabel and file conversion
• Basic usage:
obabel infile.extn –O outfile.extn

• Can also read from stdin, write to stdout, read
from a SMILES string, specify the input and
output file formats, specify conversion
options, and format specific options
– Or ask for help (obabel –H)…online docs better!

• Note: obabel has replaced the older babel

Conversion options
• Handle multimolecule files
join/m, sort, C
• Handle multicomponent molecules
r, separate
• Filter
filter, smallest/largest, s/v, f/l, unique
• Manipulate structure or atom order
addpolarh, align, b, c, canonical, d, h, gen2d/3d
• Forcefield
minimize, conformer, energy
• Conformers
readconformer, writeconformers
• Manipulate SDF properties and title
add, addfilename, addindex, addoutindex, addtotitle, append, delete,
property, title

See http://openbabel.org/docs

File-format options
• Particular file formats may have their own
specific input or output options
– To provide or handle different flavours of the file
format
– To specify additional information to include
– To provide additional functionality

• Options are listed in the help text for a format
(see next slides)

• To use:
– specify read options with –a (e.g. –ar)
– specify write options with –x (e.g. –xi)

SMILES output options

1. Add explicit Hs
Note that atom order is > obabel -:CC(=O)Cl –osmi 2. Show them in the
preserved CC(=O)Cl output
> obabel -:CC(=O)Cl –osmi –xh -h
Make atom 3 the first
[CH3]C(=O)Cl
atom… > obabel -:CC(=O)Cl -osmi -xf 3
O=C(C)Cl
…and atom 1 the > obabel -:CC(=O)Cl -osmi -xf 3 –xl 1
last O=C(Cl)C
> obabel -:CC(=O)Cl -:CC(=O)Cl -osmi -xC
ClC(=O)C Random order
O=C(Cl)C
> obabel -:CC(=O)Cl -osmi -xF "2 4"
CCl
Fragment SMILES for the fragment
composed of atoms 2 and 4

Take home message: Look through the list of options for file formats
which you frequently use (and request new options!)

Pro tip #1 “obabel –L” is your friend

Information on plugins and plugin options.

What can be done with descriptors and SDF
properties?

• Filter based on value or True/False
--filter "MW<130 & My_Property < 12"
• Sort and reverse sort --sort ~logP
• Take the N largest or smallest (or everything but)
--largest 5 MW
• Add SDF properties --add MW
• Add to title (useful for depictions) --addtotitle MW
• Remove duplicates --unique cansmi

• Create more descriptors!
– Group contribution, SMARTS descriptors or compound
descriptors are easily added via text files*

* http://open-babel.readthedocs.org/en/latest/WritePlugins/AddNewDescriptor.html

Pro Tip #2 Faster filtering

Also –aP if filtering based
on SDF properties

Pro tip #3 (Ab)use the title output format

• obabel myfile.sdf –o txt
– List the titles of all of the molecules

• obabel myfile.sdf –otxt --title “” --append MW
– List the molecular weights of all of the molecules

• obabel myfile.sdf –otxt --title “” --append
My_Property
– List the property value for all of the molecules

PNG Depiction

-xC

-xa -xt

-xu
--highlight “cCO blue”

Pro Tip #4 SVG + Firefox = User interface
• SVG has same options as PNG…
• …but drag-and-drop onto Firefox and you have
a zoomable user interface
– particularly useful for visualising multimolecule
files
– Demo showing a 1000 molecule file (only 3MB):
http://baoilleach.blogspot.co.uk/2011/06/molecular-zooming-with-open-babel-svg.html

• You could create a navigation interface for an
entire database (sponsorship opportunity!)
– E.g. make each of 1000 molecules link to another
SVG with 1000 molecules
• Multimolecule depictions can be aligned based
on substructure (also PNG)
– Demo: http://baoilleach.blogspot.co.uk/2012/02/portrait-of-molecule-as-green.html

Pro Tip #5 Automatic conversion
On Windows, create a file sdf.bat on your
Desktop with the following text:
@obabel.exe %1 –O "%~ndp1.%~n0"

If you drag-and-drop a chemical file onto
this, the file will be converted to an SDF
file.

(Rename to mol2.bat for mol2 files, etc.)

Alignment
• Open Babel does not have any code to
determine the maximum common substructure
(MCS)
– Sponsorship opportunity ahoy!

• 2D and 3D alignment is supported –align
– Based on Kabsch alignment (minimised RMSD)
– You either have to align the whole molecule
(atoms should be in same order) or else a
specified substructure (SMARTS)

• When aligning 3D structures I find it useful to --
join the results into a single structure and view
in 3D viewer (e.g. Avogadro)

Spectrophores
• Donated by Silicos-it, http://silicos-it.com/
• Usage: obspectrophore –i myfile.extn
• Requires 3D structure
– Note: it does not complain if you give it a 2D structure
– 3D conformation dependent, but orientation independent

• 48-value descriptor based on electrostatic, lipophilic and electrophilic
property values at points on a grid (or cage) and the atomic shape
deviation

Spectrophores
• Donated by Silicos-it, http://silicos-it.com/
• Usage: obspectrophore –i myfile.extn
• Requires 3D structure
– Note: it does not complain if you give it a 2D structure
– 3D conformation dependent, but orientation independent

• 48-value descriptor based on electrostatic, lipophilic and electrophilic
property values at points on a grid (or cage) and the atomic shape
deviation

• Custom code require to use spectrophores for similarity
• Silicos-it have previously trained Self-Organising Maps (SOMs) using
spectrophores for known classes of compounds and used them to
predict novel compounds for a particular class

Progamming with Open Babel
• Sometimes the GUI or command-line interface does not do
exactly what you want
– You can write your own applications or scripts

• Choice of C++, Python, Java, .NET, Perl
– But C++ and Python best supported

• Python is well-established in chemistry
– Relatively easy to learn
– Small number of commands
– Can do a lot in a few lines

• Since the full Open Babel library is quite large, to make it easy
to get started we provide a Python module Pybel
– Makes it easy to do the most common operations
– Very small number of classes and functions
– The full library is still available under-the-hood

• Google “Open Babel Python”

Using the Python Bindings
import pybel

# Read a molecule
inputfile = pybel.readfile(“mol”, “tmp.mol”)
mol = next(inputfile)

print(mol.molwt) # Show molecular weight

import pybel

# Loop over multiple molecules
inputfile = pybel.readfile(“sdf”, “tmp.sdf”)
for mol in inputfile:
# Show molecular weight
print(mol.molwt)

import pybel

if (mol.title.endswith(“_active”) and
mol.wt > 100 and “S” in mol.formula):
# Show molecular weight
print(mol.molwt)

import pybel

outputfile = pybel.Outputfile(“smi”, “tmp.smi”)
if (mol.title.endswith(“_active”) and
mol.wt > 100 and “S” in mol.formula):
# Add the molecule to the output file
outputfile.write(mol)

Learn by playing at the command-line

A cry for help
Like mailing lists?
openbabel-
discuss@lists.sf.net
Like forums?
http://forums.openbabel.org
Like to email a developer
directly?
We will ask you to email the
list :-)

Don’t forget to read the
docs first and Google it Image: Tintin44 (Flickr)

http://openbabel.org/docs

Intro to Open Babel

More Related Content

What's hot (20)

Viewers also liked (11)

Similar to Intro to Open Babel (20)

More from baoilleach (20)

Recently uploaded (20)

Intro to Open Babel

Editor's Notes