The goal is to propose a new treatment, comprising a Leishmania protein (present in the proteome of one or more of the Leishmania species) and a small molecule (or set of small molecules). Submission can be specified as:

• PDB file with a structure of a target and a bound small molecule

See the sample PDB file for download for an example of what your submission should look like.

The data for this challenge are:

• Drug targets: The set of amino acid sequences of proteins expressed and used by the pathogen organisms (Leishkmania spp.). These are the proteins that discovered drugs are to target.
• References to publicly available experimental data sets describing drug effect on drug targets (drug activity assays)
• Set of 3D structures of drug targets for structure-guided drug discovery.
• Available structures for drug targets (experimental and predicted/theoretical) will be available in Google Storage repository.
• Repository will be continuously updated with new models, when they become available.
• References to freely available, online resources to produce accurate 3D models of proteins not available in the Google Storage repository
• Set of commercially available, approved drug molecules (in SMILES/SDF/MOL2/PDB formats), together with their ChEMBL and PubChem identifiers to cross-reference with drug activity assays
• Set of experimentally tested, safe for humans drug-like molecules (in SMILES/SDF/MOL2/PDB formats)
• References to publicly available resources enabling discovery of new drug targets
• Protein-protein interaction databases (e.g. STRING)
• Metabolic pathway data (e.g. BioCyc, KEGG)
• References to freely available cheminformatics software, their reference material and worked-out code examples, enabling in particular:
• Computing similarity between different drug targets
• Computing fixed- and variable-length representations of drug candidates
• Converting small molecule (drug candidate) representations between formats
• Producing a set of realistic conformers (3D configurations) for a molecule, as defined by a molecular topology graph (i.e. atoms and bonds)
• Computing the binding affinity of a protein to a given target in silico through molecular docking

Submission definition

• Single submission comprises a single PDB file (pose) containing the structure (or part of it) of a protein on a drug target list and at least one drug candidate
• If a submitted pose contains a drug candidate bound to drug target (small molecule docked into the protein), we will take this pose into consideration when evaluating the score
• If the submitted pose contains drug candidate and drug target, but they are not interacting (unbound pose), we will perform a standard docking protocol internally, to identify the most favorable binding pose.
• If the submitted pose contains more than one small molecule, we will dock them to the submitted protein simultaneously. If you want to submit more than one protein, or more than one binding pose of the same protein, do multiple submissions
• If the submitted pose contains no small molecule, no protein chain, or protein chain submitted is not on a drug target list, nor is a subset of a chain on a drug target list, it will not be evaluated.
• PDB format is the only format accepted. Every atom designated as ‘ATOM’ will be treated as belonging to the drug target. Every atom designated as ‘HETATM’ will be treated as belonging to the drug candidate.
• Submissions in other formats will not be accepted.
• We recommend RDkit or OpenBabel (both freely available, open source software) for format interconversion.
• The geometry of ligand (drug candidate) does not need to be optimized
• We recommend that drug candidate is located in the binding pocket, but it is not necessary for successful submission.
• We recommend RosettaLigand, AutoDock 4, orAutodock Vina for docking.