Roadmap#

Deemian is still in its early phase and thus only has minimum features. However, there are many features planned, including:

  • Basic interactions: Hydrophobic, hydrogen bond, electrostatic interaction, aromatic interaction.

  • Other interactions: Polar interaction (loose H-bond), pi-cation interaction, halogen bond, carbonyl interaction, metal complex/coordination, etc.

  • Bridged interactions: Water-bridged interaction and metal-bridged interaction.

  • Anti-interactions: Steric clash interaction, same-charge interaction, polar-non polar interaction.

  • Element and atom type interactions: Calculate the interaction between elements or Sybil atom type (Ballester, Schreyer, and Blundel, 2014).

  • Bitstring AKA Bitvector: The ability to present interaction as bitstring which allows for interaction comparison. Also add the ability to calculate the interaction similarity.

  • Better compatibility with Docking results: Create commands to make virtual screening result analysis easier.

  • Better compatibility with MD results: Create commands to make MD result analysis easier.

  • Interaction Pseudo-Atom (IPA): Present interaction as IPA (Desaphy et.al., 2013) with three kind of placement, at first subject, at the middle of interaction, and at second subject.

  • Hashing capability: Integrating hashing and scoring to enable TIFP (Desaphy et. al., 2013) and PLEC (Wójcikowski et. al., 2018).

  • Structure alignment: Add the capability to align 3D structure. Useful for accumulating IPA from different protein-ligand complexes, protein family (e.g. Kinase family) comparison, virus mutation study.

  • Group/substructure interaction: Allows interaction analysis between chemical groups or large substructure.

  • Interacting surface area: Calculate the interacting surface area for each interaction type.

  • Integration with Jupyter: Allows the Deemian data to be processed and analyzed within Jupyter Notebook/Lab. Also allows visualization with NGLView.

  • Extension with Python: Allows user to create custom instruction via Python scripting.

  • Support various file format: Theoretically any file format that is supported by RDKit and NGL can be loaded.

  • Support CGMD results: especially Martini 3, which gain a lot of traction lately and useful in large scale simulation.