Why biobuild?

biobuild is a mocule building suite for cheminformatics and bioinformatics. It is designed to be a simple, easy to use, and extensible tool for building larger molecular structures from smaller components. biobuild tries to leverage the power of geometric operations in combination with machine learning algorithms to facilitate the building of complex molecular structures where direct imputation from SMILES may not be feasible.

In order to facilitate the building of larger structures, biobuild provides a set of tools to not only easily and quickly connect smaller molecules to larger structures, but also integrates a extensible database of available molecules that can be loaded without the need to download external data sources, or read files from disk.

Why “bio”-build?

biobuild is designed on top of biopython and was originally designed to build larger biomolecules, specifically glycans. However, the underlying algorithms and implementations are not limited to these glycobiology and hence, biobuild was created as a stand-along framework for molecule building. biopython was chosen as backend due to its great flexibility when it comes to adding and removing atoms to and from structures. However, biobuild is not limited to biopython but exports import and export of structures to and from rdkit, another widely used cheminformatics library.

What can biobuild do?

biobuild can be used to build larger molecules from smaller components - that’s pretty much it. However, given its integrated compounds database, it can also be used to quickly obtain molecular structures for other purposes such as docking. Its ability to convert to and from rdkit also makes it a great tool to edit rdkit structures in a more intuitive way. Lastly, biobuild offers a suite of conformational optimization algorithms to improve the geometry of structures. Hence, biobuild can be used to sample conformers for molecular docking or the generation of a deep learning dataset.

What can biobuild not do?

biobuild is not a quantum chemistry package. It does not offer any quantum chemical calculations, nor does it offer any tools to analyze the electronic structure of molecules (aside from whatever is available from biopython, of course). It is also not a molecular dynamics package. Also, currently biobuild is limited to creating single bond-connections between molecules. Hence, biobuild cannot create (small) ring structures. However, using fragments with the desired ring structures already present (e.g. from pubchem) large circular molecules can be built regardless. However, biobuild is currently not optimized for circular molecules of any kind, hence, users wishing to create circular molecules will likely want to use molecular dynamics to optimize the geometry of any structure they create. Finally, biobuild does not generate molecules for you! It is not a de novo molecule generator that tries to find molecules with certain properties - e.g. molecules that bind a certain protein pocket. The user needs to know what they want to build and have some idea of the steps involved - biobuild will help them build it, that’s the deal.

Who is biobuild for?

Anyone who would like to build larger molecules from smaller components. biobuild is designed to be easy to use and intuitive. Most of its user-relevant functions are accessible through toplevel functions, or methods of the Molecule class. That in mind, biobuild is designed to work interactively with the user, for instance in a Jupyter notebook - biobuild offers some customizable 3D visualization. However, biobuild is equally suited for scripting and integration into larger workflows such as snakemake pipelines. How it is used is up to you!