I have shared with scientists three solutions, step1, step2, step3 that show a structure from origin (step1), trim, microidealize which yielded 2 point gain on the trimmed section (step2) and then the surprise: (step3) a one thousand sevenhundred point loss because of those 2 point gain, which could not be recovered. Only a restore credit best brought me back to the original score.
I find it unthinkable that a small 2 point gain could cause a 1.700 point loss without any new clashes. Seems a scoring anomaly to me.
Hi @"Bletchley Park" just wanted to clarify something- is the issue that you're describing a score decrease upon untrimming? If so this is a possible side effect when using the trim tool if coordinates of residues outside of the trimmed region are not taken into account.
@apetrides Yes it is, but as I said "without any new clashes". Can you elaborate on how this 'side effect' could cause such a big drop in score ?
Hi @"Bletchley Park" thanks again for sending over these example solutions. We ran some analysis and found that the result of the score drop was the cart bonded score term. We suspect that this had something to do with the fact that our previous iteration of the trim tool included additional residues to help lock the backbone in place. We are already about to push a change that would prevent these additions from spilling across chain breaks. This is undoubtedly related to the issue you mentioned here.
"We are already about to push a change that would prevent these additions from spilling across chain breaks" That makes me very happy ! Thank you @apetrides
Should be fixed in the latest DevPrev @"Bletchley Park"
