RDKit

RDKit is an advanced and complex software solution which includes several different tools and utilities meant to be used in cheminformatics and machine learning operations, evidently being created with experienced users in mind.

The toolkit is open-source, released under the BSD license, meaning it can be used as well as developed by anyone. The core data structures, as well as the algorithms comprised by RDKit, are all written in C++.

It relies on a Python 2.7 wrapper that was developed using Boost.Python, but it also uses Java and C# wrappers that were created with SWIG. RDKit can be used for both two and three-dimensional operations, functioning as a description generation instrument for machine learning tasks. Moreover, the molecular database cartridge is created for PostgreSQL while also featuring cheminformatics nodes for KNIME.

In terms of functionality, there are several input and output formats supported by RDKit, for instance, SMILES/SMARTS, SDF, TDT, SLN, Corina Mol2, PDB, and others. The cheminformatics side of the toolkit offers substructure searching, canonical SMILES, chemical reactions, molecular serialization, chirality supports and chemical transformations.

The molecular description library provides topological, compositional and electrotopological states as well as feature-map vectors, whereas the machine learning side of the software can handle clustering and information theory.

The installation process has three main prerequisites, namely Python 2.7, Numpy and PIL, the latter being a library. Other possibly useful software is aggdraw, matplotlib, ipython, and win32all. RDKit is a multi-platform solution, but installation is platform specific, so users can refer to the hefty documentation that it accompanies it to find the one that applies to their situation.