I made a list of my questions for scientists- origami, knots, MRSA, and others

Started by jawz101

jawz101 Lv 1

I'm embarrassed to ask these brainfarts but here they are:

Have there been any challenges to test MRSA treatments?

Are loops less significant than helices and sheets? Would a design based solely on loops do anything in real world?

If you have 2 sheets and you bond them, would it be better if they bond flat, idealize first, or idealize SS, or idealize + idealize SS and then bond those idealized sheets together?

I've wondered if the way scoring works is based on the number of manipulations it would take to achieve the end shape. Is that how energy is scored?
Like if I handed you a coat hanger that I had bent up into a certain shape, is the goal to see what the minimum amount of manipulations it takes to match that shape identically?

I wonder if the goal is simple, repeatable design to achieve the best outcome.
To that end, is the goal to to find a design that is easily reproducible since a drug would have to be mass produced?

One of my first feelings when playing the game was that quote attributed to Michalangelo.

Something like “The sculpture is already complete within the marble block, before I start my work. It is already there, I just have to chisel away the superfluous material.”

What I wondered: is there an approach whereby a perfect shape is known but it's a matter of making that shape with the lowest energy.

That is, could an approach be to reverse engineer what the manipulations would be to make a known, desired shape?

Does protein folding have any relation to origami or knot tying? Maybe 20 years ago I had read that the mathematics of origami was a newer interest in math. Would anything from either of these things translate much to protein folding?

There is probably better info than a couple of wikpedia pages on these 2 math things. I just wanted to find something that talked about the ideas.

beta_helix Staff Lv 1

The first paper I ever published was about knots in protein structure prediction: https://academic.oup.com/bioinformatics/article/22/14/e252/228283
Direct link to PDF here: http://www.cis.umassd.edu/~fkhatib/Papers/Knotfind.pdf

tl;dr Rosetta and other automated programs would artificially create knotted predictions, so we needed to come up with a filter to detect knots quickly (and eventually fix the algorithms from creating so many knots in the first place!)

donuts554 Lv 1

I don't think you should be that embarrassed to say your thoughts here.

I don't know if there have been any challenges to test MRSA treatments or not.

I think that loops are less significant than helices and sheets because the backbone atoms in loops don't make much hydrogen bonds. However, I personally think that a design solely based on loops would do something useful in the real world, because of the bonding between the polar hydrophilic sidechains making up for the bonding between backbone atoms.

I don't think that scoring is based on the number of manipulations it would take to achieve the end shape. I also don't think that the goal is to find a design that is easily reproducible, even though a drug has to be mass produced. It is because energy is just a measurement that is related to the forces between atoms in a amino acid chain, not by the simplicity of the folding sequence.
I think that what you are talking about is more about entropy, as simple small folds to make a structure do seem easier to make than more complicated ones. But, there are a large amount of complex things and proteins in the world, so I think designing proteins should be more about structure and function . Any normal-sized physiologically stable protein can be mass produced by using yeast and/or ribosomes, etc.

I think that what you are saying about Michelangelo's quote is somewhat true. I think protein structure prediction from a given amino acid sequence is basically what you are saying, so I think there is an approach where a perfect shape is known but that it is a matter of making that shape with the lowest energy.

I also have wondered whether protein folding has any relation to origami or knot tying, and I am currently investigating not whether, but what are some of the relations between origami and protein folding.

From,
donuts554

jawz101 Lv 1

Awesome responses from you both! I am very happy these questions were fun to think about. I've been playing with Foldit almost every day for the past 6 months or so and after a while those funny questions build up.

Formula350 Lv 1

As the saying goes, "The only stupid question, is the one not asked!"

The usual disclaimer: I Am Not A Scientist
I have a very crude and limited understanding of BioChem, and what little I do know, I have Foldit to thank.

That being said…
Are loops less significant than helices and sheets? Would a design based solely on loops do anything in real world?
Just based on my time playing various puzzles in Foldit, I'd say that there must be…
Which after a quick google search… I am pleased to be able to tell you that YES… there are such things are pure-loop proteins. Not just "that fold" but which are found all about in nature! :)
https://www.google.com/search?q=structureless+proteins+in+nature

If you have 2 sheets and you bond them, would it be better if they bond flat, idealize first, or idealize SS, or idealize + idealize SS and then bond those idealized sheets together?
This isn't a super helpful reply but, I imagine that how you bond them isn't really going to matter too much, just as long as they're bonded in the correct way that they WANT to be. I'm definitely not the person to ask, but "Chiralty" is a protein term that might help shed some light on this.

Now as far as Foldit goes… Idealize SS automatically does an Idealize of the segments you use the tool on.
You're going to have a much easier time getting them to bond together if you first Idealize SS, to give them that slight twist, which inherently gives the sheets a 'curve' as they get longer or as you bond more together.

That curve is, I believe, rather necessary in order to allow there to be sufficient room for neighboring Sidechains. While some of them are very happy to bond with another sidechain, that's only at their tips. The rest of their… uh…. "torso" (? lol) wants it own personal space. That's right, even proteins obey Social Distancing!
Crude ASCII Representation:
If | = a Sidechain
Then in order to pack them together on a flat Sheet, they'd end up being like this…
| | |
That may be too far for the Sheets to bond, therefore being very much NOT a sheet :(

BUT if you give the Sheets just a slight curve, now you can pack them closer together…
\|/
Gives their tops enough room to breath, but at their base, the Sheet, it's close enough for a strong bond! :D


Hope that helps a tiny bit :)
-Form

joshmiller Staff Lv 1

Great thread jawz, thanks for asking. Disclaimer: I'm not a scientist, take this response with a grain of salt.

> I'm embarrassed to ask these brainfarts but here they are:
Don't be embarrassed, these are good questions.

>Are loops less significant than helices and sheets? Would a design based solely on loops do anything in real world?
Helices and sheets are just loops bended in a particular way so as to form more hydrogen bonds. There can exist loops that form plenty of hydrogen bonds without these configurations. To me, it is analogous to something like "Can you build a building without triangles?" Yes, probably, but these common design patterns exist because they're very stable and easy to make.

> If you have 2 sheets and you bond them, would it be better if they bond flat, idealize first, or idealize SS, or idealize + idealize SS and then bond those idealized sheets together?
This is a Foldit-specific problem. Personally, I find it easier to bond sheets together by laying them out flat (ideal/SS) first, then bonding. In natural proteins, secondary structures (single strands) tend to fold up before tertiary structures (strands coming together to be sheets), because local attractive forces (like backbone h-bonds with nearby residues) act before global attractive forces (like sidechain h-bonds with other strands). The importannt point to remember, though, is that they typically don't stay flat. Natural sheets are bendy or curvy, usually.

> I've wondered if the way scoring works is based on the number of manipulations it would take to achieve the end shape. Is that how energy is scored?
Not at all, and this is a good question for clarification! The score function is based purely on the energy of the protein. That is, we run a lot of calculations on how much every atom likes where it is based on where every other atom is, and that's your score. There's no consideration to how you got there, or how much manipulation it took. However, this brings up another good point that sometimes players "over-optimize" a solution – our calculations are only an approximation, and sometimes if you squeeze out too many final points, you end up making the protein less realistic because you're optimizing for the quirks of the approximations. This is why we introduced move limits in Sketchbook puzzles, and we will likely bring those back when we need to limit the amount of manipulation for avoiding "over-optimizing".

> I wonder if the goal is simple, repeatable design to achieve the best outcome.

To that end, is the goal to to find a design that is easily reproducible since a drug would have to be mass produced?</b>

Yes, sort of. "Repeatable design" in this sense means it has a smooth energy landscape. Reproducible, then, is a function of the protein and its design, not of humans making it.

> is there an approach whereby a perfect shape is known but it's a matter of making that shape with the lowest energy.

That is, could an approach be to reverse engineer what the manipulations would be to make a known, desired shape?</b>
The point is somewhat moot, I think, because the energy is directly derived from the shape. You can't make the same shape with different energies - if they have different energies, they are different shapes. I think a more apt question would be, suppose we want the protein to take shape A, how do we make sure it doesn't take shape B or C? This relates, again, to the energy landscape blog post I've linked twice already, haha.

Great questions, again.
Happy to talk about this further on the Foldit Discord.

joshmiller Staff Lv 1

Oh! Forgot to answer your last question. It's a very neat perspective. It's a really tangential connection, I think, but there might be some value to that analogy. I'd have to think on it more. You might have to ask some more veteran players what they think.

For me, I think of it like this: hydrophobics are sticky and gluey, attaching to anything. Hydrogen bonds are like a small patch of velcro, they make good connections with each other, but as a whole aren't very strong. Together, the hydrophobics do the bulk of the folding and the h-bonds ensure that it folds in the right way by snapping into place (as opposed to the many other ways glue could stick stuff together). Not sure if that analogy helps at all.

jeff101 Lv 1

> Is there an approach whereby a perfect shape is known but it's a matter of making that shape with the lowest energy. > That is, could an approach be to reverse engineer what the manipulations would be to make a known, desired shape?

Sometimes in Design Puzzles, I use many bands to enforce a certain shape,
then I use mutate to find different amino acid sequences that score better
(have lower energy) in that shape. Different amino acids definitely have
different likelihoods of being involved in secondary structures like
helices (H), sheets (E), and loops (L). If you don't like bands, the 
Blueprint Tool has yellow boxes that enforce certain constraints on loops. 
You can keep these yellow boxes around while you mutate to find sequences 
more likely to fold into the loop's desired shape. You can also set the 
desired secondary structure (H E or L) for certain parts of the protein, 
then in the Selection Interface, select these parts and then press the 
buttons 5 then 2 then 5 then 2. This will make the selected helix (H) 
and sheet (E) parts look like ideal helices and sheets. After that, you 
can run mutate, shake, or wiggle sidechains, and the sequence and score 
will change, but the overall shape will remain the same.

The Forum post below is about using bands to force the protein into a 
square barrel shape composed of many parallel beta-strands separated
by loops, then using mutate to find a sequence more likely to form 
this shape in nature. I hope this post helps.
 
https://fold.it/portal/node/2005553

Good Luck, and Happy Folding!