Based on Fang-Yu's comments, there's significant delay 1) loading the first molecule (up to 10 minutes :( !), and 2) getting the next molecule from the list (~11 seconds). This is happening because of some silly implementation decisions I made when working with a small set of molecules, and there's definitely a way to speed this up. My first thought is to pre-populate the molecule queue and "stage" the next few molecules, so that the user isn't stuck waiting while they're being retrieved.