Hacker News new | comments | show | ask | jobs | submit login

This is going to be incredibly big for the protein structure predictors as well. Most of the best models incorporate bond-breaking and conformational switching moves as described in the function of this protein, but the vast combinatorial possibilities make it hard to resolve what the folding paths are to the global minimum and which are "unnatural" and leave you sitting in a local minima. Tweaking their models with real-world input from this type of imaging and more sophisticated statistics about the frequency of moves will work wonders for their automated efforts.

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | DMCA | Apply to YC | Contact