Recipe: Bravo Cut and Close

created by bravosk8erboy

Profile

• Name: Bravo Cut and Close
• ID: 109020
• Shared with: Public
• Parent: None
• Children:
  • Trim and Un-Trim scoring bug
  • Test recipe 09 (Close cuts with trim)
• Created on: April 11, 2025 at 17:39 PM UTC
• Updated on: April 18, 2025 at 19:45 PM UTC
• Description:

  idea credit to Serca. ; just cuts and closes cut

Code

-- this recipe just cuts and closes that cut and checks if the score improves.
-- seems to move the oxygen atom on the backbone of the lower numbered segment to improve the score.
-- credit to Serca for this discovery.
-- i recommend runing this before any recipe that involves cuts, idealizeSS, idealize peptide bonds, remixes, or rebuilds.
-- i recommend runing this after any rebuild or remix recipe.
-- there is a chance that running this might make it harder to rebuild or remix because those will probably have bad Oxygen bonds after scoring lower.
-- in that case you might want to only run this at the end with other end game recipes
---------------------------------------------------------------------------------------------------------------------------------------
-- experimental review by Bravo
-- cutting and closing apears to slightly move the oxygen atom on the backbone of the lower numbered segment of the cut.
-- example: if you make a cut and close it between segments 21 and 22. the oxygen atom (atom 4) on segment 21 will slightly move.
    -- (the oxygen on the carboxyl side of the cut)
-- this doesn't always improve the score.
-- cuts and closes seem to affect the score of nearby segments. 
-- different subscores will change when segements are locally located near each other.
    -- (example: if segment 5 and 105 are close enough in space the subscores will change. 
    --  the closer the segments the more subscore changes will occur)
-- this cut and close method seems to improve the score of proteins that have been fully wiggled on low or high.
-- this cut and close method also seems to improve score outside of any form of shaking or picking sidechains.
-- there is a chance that rebuild and remix fixes this but with the drastic shape change it is hard to tell.
    -- Rebuild:
    -- tested rebuilding the whole protein then running this recipe and got a 112.6 point gain across 80 segments.
    -- after an entire rebuild of the same protein this produced an additional gain of 67.8 points, suggesting it should be run after every rebuild.
    -- Remix:
    -- tested remixing the whole protein then running this recipe and got a 499.1 point gain across 80 segments.
    -- after an entire rebuild of the same protein this produced an additional gain of 172.9 points, suggesting it should be run after every remix.
-- idealize secondary struture "Sheet" will fix this. idealizeSS the whole protein and then run this recipe and only got a 0.0000000028 point gain for the whole protein.
-- idealize secondary struture "Helix" will fix this. idealizeSS the whole protein and then run this recipe and only got a 0.0000000092 point gain for the whole protein.
-- idealize secondary struture "Loop"  will fix this. idealizeSS the whole protein and then run this recipe and got no point gain for the whole protein.
-- idealize peptide bonds "Sheet" will fix this. idealizePB the whole protein and then run this recipe and got no point gain for the whole protein.
-- idealize peptide bonds "Helix" will fix this. idealizePB the whole protein and then run this recipe and got no point gain for the whole protein.
-- idealize peptide bonds "Loop"  will fix this. idealizePB the whole protein and then run this recipe and got no point gain for the whole protein.

-- cut, rebuild, remix, or idealize recipes could have the chance of fixing this but making a worse shape, overall.
-- this could give a score improvement less then cut and close alone.

---------------------------------------------------------------------------------------------------------------------------------------
--code writen by microsoft copilot


-- Initialize variables
local starting_score = current.GetScore()
local best_score = starting_score
local segment_count = structure.GetCount()
save.Quicksave(1)

-- Function to check and update the best score
function check_and_update_score(segment)
    local new_score = current.GetScore()
    if new_score > best_score then
        gain = new_score - best_score
        best_score = new_score
        save.Quicksave(1)
        print(" Seg: " .. segment .. " Score: " .. best_score .. "   Gain: " .. gain)
        return true -- Score improved
    else
        save.Quickload(1)
        return false -- Score did not improve
    end
end

-- Main function
function main()
    local score_improved = true
    
    while score_improved do
        score_improved = false
        -- Print the starting score
        print("Starting score: " .. starting_score)
        
        -- Loop through each segment
        for i = 1, segment_count do
            -- Apply a cut to the segment
            structure.InsertCut(i)
            
            -- Close the cut
            structure.DeleteCut(i)
            
            -- Check and update the score
            if check_and_update_score(i) then
                score_improved = true
            end
        end
    end

    -- Print the best score and total gain
    ending_print()
end

function ending_print()
    -- Print the best score and total gain
    local total_gain = best_score - starting_score
    print("Best score: " .. best_score)
    print("Total gain: " .. total_gain)
    
end

-- Error handler function
function error_handler(err)
    print("Error: " .. err)
    check_and_update_score(0)
    ending_print()
end

-- Execute main function with error handling
xpcall(main, error_handler)

Comments

[LociOiling]

The same general idea has been around for a while, although usually there's some wiggling involved.

See TvdL Be careful with that axe Eugene, which was originally on punch cards.

[bravosk8erboy]

I understand there are a lot of different fuse/wiggle with cuts recipes. this one doesn't do any of that. I think it shows that the wiggle involved in those recipes might not be where the gain comes from. sometimes it's the cut and close alone that gives the gain. just sharing this incase anyone might want to explore the idea with other recipes. I think there could be value added into rebuild / remix recipes to cut and close also when fuzing. after rebuild or remixes I have seen up to 500 points gain on a segment from this that a rebuild left available.

[Serca]

Loci, the main difference with this recipe is that we don’t do any Wiggle or Shake at all. We just make a cut and then immediately close it - and we get a score boost. What’s most surprising to me is that after doing this, Wiggle at the end often doesn’t add any more points. It’s like we’ve significantly increased the score without actually changing the protein’s conformation. Bravo found out that the score change probably comes from the shift in the oxygen atom’s position. But I still can’t understand why there could be a case where Wiggle doesn’t change the protein’s shape after a major score jump from the cut & close.

[Serca]

I’d say this cut & close recipe, even though it cuts the backbone, behaves more like the Shake tool. You could say it does a "Shake" in a spot where neither Shake nor Wiggle can reach - and even Rebuild isn’t very effective there.