Fragment-based drug discovery is a method to design high-affinity selective compounds starting from small fragments, e.g. by splitting the receptor pocket into different sub-pockets and optimizing fragment structures in each of them before connecting the fragments together.
Popular approaches include the following:
- fragment growing
- attaching additional functional groups onto the fragment to make it bigger, hopefully increasing the contact surface and improve interaction.
- fragment merging
- allow the sub-pockets of fragment optimization overlap each other and combine fragments found in the pocket on the common substructure.
- fragment linking
- optimize the fragments in adjecent sub-pockets and connect them with a linker.
In Foldit, the fragment growing approach is rather straight-forward. It is almost always what a player does.
Fragment merging and linking, on the other hand, is not very efficient in Foldit. A library search for each part of the structure would help to make the compound library more useful for these two approaches. Alternatively, options like substructure search or superstructure search may also helps.
Currently it is easier to be done by searching the target fragments on web in external database, then build them manually in the game and connect the different fragments. It would be nice if the whole process can be done within the game!
Fragment merging and linking is not build in Foldit, is it ?
Merging and linking not directly supported in Foldit as I'm awared of. Not sure if there's any tricks to do it.
Can somehow be done separately, e.g. use different tracks to optimize fragments then manually build the merged or linked structure.
Or select fragments out of the game and build them in the game and optimize, then test with linkers or central substructure etc, which is what I'm trying in this puzzle.
With the core filter in puzzle 2313 cannot use the in game compound library to search for compounds of the 2nd fragment.
So this time since I gave myself a rule to use the linking approach, I did the following:
- design the first part (starting with the same design as puzzle 2310 with MW limit 300),
- test where I put the linker (actually I had it in the previous design),
- add another fragment inspired by external library to the linker.
Actually it's a combination of fragment growing and (manual) linking.
A downside is that linkers can be very polar (not all, but many common linkers are quite polar. There are also many carbon-based and non-polar linkers, just hard to handle for me). After adding the linker can be difficult to add more h-doners and acceptors while staying within the TPSA limit.
Fragment merging and linking is something we've thought about adding (and on the list to support), but it needs some technical infrastructure updates in the client to accommodate properly.
