I consider myself sufficiently expert on the topic to review. I really like the idea of the proposal but think it needs more polishing before it would be worthy of funding. A challenge is that I am sympathetic to the approach being proposed but a typical HHMI reviewer may not be. More work is needed to get across a strong message of how novel, ambitious, and game-changing this approach can be. You need to explain the problem that you are trying to solve (identifying potential new drug targets for bcr-abl inhibitor resistant disease from the vast knowledge trapped and unsynthesized in the biomedical literature). Then convince the reviewer that your hybrid crowd-sourcing + ML approach is an exciting and novel new way to potentially solve this problem.
Thanks for your feedback. I found it difficult to balance the need to provide background information for the reviewers with the need to explain why the technique is novel. The strict three page word limit is preventing me from going into the detail I would ideally want.
Did you have any general recommendations? I also agree that the proposal needs more polishing.