[octet-devel] Modular AtomQuery, MoleculeQuery support now available in Octet.

[rich a. <che...@ya...>]

I have just committed changes to the Octet CVS that
address the issue of modular AtomQuery and
MoleculeQuery support (http://octet.sourceforge.net).
Most of the changes can be found in the
net.sourceforge.octet.query package
(http://cvs.sourceforge.net/viewcvs.py/octet/octet/src/net/sourceforge/octet/query/)

Examples of AtomQueries now available include:

MoleculeFeatureAtomQuery - Atom membership in features
perceived by a MoleculeQuery (aromaticity, ring
membership, etc.)

AtomicNumberAtomQuery - atomic number

ImplicitHydrogenAtomQuery - implicit H count

NegatingAtomQuery - negates the result of another
AtomQuery

NeighborCountAtomQuery - neighbor count (excludes
implicit hydrogens)

TemplateAtomQuery - matches atoms corresponding to an
Atom template

WildcardAtomQuery - matches all Atoms

More are on the way.

The construction of composite AtomQueries that can
contain nested boolean logic is now supported. These
AtomQueries can either be used to directly match
individual Atoms in a Molecule, or as part of a larger
MoleculeQuery created, for example, with a
MoleculeQueryBuilder.

Examples of MoleculeQueries now available include:

AllCyclesMoleculeQuery - perceives cyclic features

AromaticityQuery - perceives aromatic features

CycleSystemQuery - perceives cycle systems (i.e. fused
ring systems useful in SDG, aromaticity perception,
and structure->name schemes).

To improve MoleculeQuery performance, a
CachedMoleculeQuery is now available that will cache
the results of a query and automatically use the
cached result when appropriate. Only one cached result
is now supported, but this could easily be extended to
enable any number of results to be cached by a
MoleculeQuery - without the need to rewrite any
existing MoleculeQueries (Decorator Pattern).

To make it easier to use this new capability, work to
create a SmartsQueryFactory is underway. The idea is
to allow clients to supply the factory with a SMARTS
atom primitive as a String and get a pre-configured
AtomQuery in return. Currently "a", "A", and "*" are
supported. Extension to other query ontologies is, of
course, also possible.

Many of these changes are intended to move toward an
object-oriented Atom typing system. SMARTS is one
apporoach, but I believe that the object-oriented
approach is much more flexible because clients can use
a language that is infinitely more expressive than
SMARTS - Java. Atom typing is used extensively in
QSAR/QSPR to calculate Molecular properties (TPSA,
etc.)

JUnit tests for this functionality are available
(net.sourceforge.octet.junit).

This code will still need some debugging and
refactoring, but I believe the overall design may be a
useful starting point.

best,
rich


		
__________________________________ 
Do you Yahoo!? 
Jazz up your holiday email with celebrity designs. Learn more. 
http://celebrity.mail.yahoo.com