Structure Indexer

This is a self-contained structure indexer that uses Lucene as the underlying storage and indexing engine. The indexing architecture is based on an inverted indexing technique developed within NCATS. Please consult the wiki (coming soon) for additional details.

Contact: [email protected]

Building

If you're building for the first time, you'll need to add the jchem.jar to your local maven repository. To do this, execute the following command:

mvn install:install-file \
  -Dfile=lib/jchem.jar \
  -DgroupId=chemaxon \
  -DartifactId=jchem \
  -Dversion=3.2.12 \
  -Dpackaging=jar \
  -DgeneratePom=true

Then simply do mvn package to build the structure-indexer jar file. The bin directory contains the following wrapper scripts:

indexer is the main driver that is used to build the index. See indexer -h for complete usage. Here is an running example:

indexer index_dir BindingDB2D.sdf

searcher is the client driver that provides a command-line interface for searching and filtering. See searcher -h for complete usage. For example, consider the following command:

searcher  -fsmiles -F_natoms=20:22 -F_molwt=280.:300. -F_source=BindingDB2D -s sim -t.9 idx "N1c2ccccc2NC(=O)c2cccnc12"

This example performs similarity searching against the index idx for the given structure with a Tanimoto cutoff of 0.9, number of atoms in the range [20,22], molecular weight in the range [280, 300], only from the source BindingDB2D, and outputs the matches as SMILES format.