rosie4loop Lv 1
The smooth and interactive interface in Foldit and its real-time energy evaluation making it a promising tool for teaching biochemistry, with the comprehensive in-game tutorial on protein structure. The ligand design feature also offers support for teaching structure-based drug design for researchers from other fields, or for an undergraduate pharmaceutical chemistry course. With only a minimal number of buttons in the game, the learning curve would not be as steep as major modelling software that's freely available.
However there are several points to be improved for a better support of the education in drug design. The lack of explanation on objective filters is rather critical, which require the student or educator to rely on external source to fully understand the reason behind. There are some information on Foldit website in the comments of each puzzle and in the forum (e.g. this post is more detailed but still missing something ). The currently provided information mainly tells "what is it" but not much on "why we need this", "why this value as cutoff". In addition, it's always better if the information can be found in the game itself. The community-contributed Wiki have more information on these objectives, but still may not explain "why".
For educators intended to have a small molecule puzzle that last longer than a week, the requirement of command line script in Rosetta for preparing ligand topology and the complications of setting objectives is also a barrier for those without computational background.
Hence I propose a sandbox small-molecule design puzzle for educational purpose. It can be in the form like the protein design sandbox so other players can also compete here, or in the educational mode. With such a puzzle, educators don't need to learn to setup a puzzle nor need to worry about the constantly changing objectives in different rounds of puzzle. I wish this puzzle would have the following features:
- all objective filters commonly used in drug design are included, with the value set to be either drug-like or lead-like (with explanation on why choosing the value).
- information (i) mark for popup explanation for each objective filter. If it's too long to include in the game interface, add a link like in the insulin educational puzzle. It better be something separated from the info you get when hovering the cursor on the filter since the current popup would change when you have satisfied the objective.
Explanation of each filter should include the following:
- What is it. If it's in abbreviations, state the full name of the filter. This is already done in the comment of each puzzle, but not in game (better be something separated from the current popup when hovering mouse on filter in game).
- Why we need this filter. In layman's terms. Don't explain with equations. There is always a reason to adapt an objective filter. For example,
- metrics like no. of donor/acceptor, molecular weight, cLogP, TPSA as a popular indicator of drug-like compounds. These values are summarized from the analysis of approved orally available drugs. These metrics are related to the drug's "journey in human body". We want it to be able to pass the barriers in our body, e.g. cell membrane, transport in the bloodstream, etc, before it reach the drug target.
- Number of rotatable bonds is related to whether a ligand wants to be trapped in target pocket or be free in water. In some cases it may also affect the selectivity and specificity.
- Bad group filters with substructure labelled by PAINS/BRENK etc and others can be e.g.
- common structures of false-positive hits, i.e. this structure like to pretend its a good drug in experiments.
- structures that are unstable and would convert to other form in water
- structures that are unstable and would breakdown in water
- structures that would react and form covalent bond with one or more protein and may lead to toxicity. This can be a strategy in drug design, but if we aim a competitive drug we want to avoid this, particularly if its too reactive and have non-specific reaction. Cysteines are popular to be attacked by reactive functional groups, residues with hydroxyl like serine are also commonly targeted, as well as lysines.
I really look forward for a day when Foldit could replace 80% of PowerPoint and worksheet in biochemistry/pharmaceutical chemistry education!