/sci/ - Science & Math » Thread #11066934

50KiB, 450x439, random protein.jpg

View Same Google iqdb SauceNAO

Protein modelling/simulation software

Anonymous Thu 17 Oct 12:22:18 2019 No.11066934 View Reply Original Report

Quoted By: >>11067120 >>11067744 >>11069169 >>11070780 >>11071041

What's the best / your favourite protein modelling software? Looking for something that can resolve the structure from aa sequence and simulate protein - protein and protein - substrate interactions.

I'm looking for something in Python and I heard about PyMol, The Molecular Modelling Toolkit and PyRosetta. Are these good? What does /sci/ use?

Anonymous

Anonymous Thu 17 Oct 2019 12:42:54 No.11066954 Report

Quoted By: >>11067137

what is used protein modelling for?

Anonymous

Anonymous Thu 17 Oct 2019 14:10:35 No.11067120 Report

Quoted By: >>11067875

>>11066934
I hope you have a supercomputer anon because you're going to need it.

Anonymous

View Same Google iqdb SauceNAO Screenshot_20191017-102642_YouTube.jpg, 734KiB, 2220x1080

Anonymous Thu 17 Oct 2019 14:27:09 No.11067137 Report

Quoted By: >>11067604

>>11066954
Biochemistry, molecular biology, virology, pharmaceutical development, synthetic biology, biomaterials/polymer science, pathology, etc.

Anonymous

Anonymous Thu 17 Oct 2019 17:58:43 No.11067604 Report

Quoted By: >>11067655

>>11067137
it's not my field of study... but can you tell me why you would need to model the behavior of proteins? And why does this need a super computer?

Anonymous

Anonymous Thu 17 Oct 2019 18:20:54 No.11067655 Report

Quoted By: >>11067684

>>11067604
Protiens are very important macromolecules for all living things. They are very complex and understanding how they work and interact is crucial to understanding human disease and huge portions of biochemistry and biology. The reason you would need a super computer is because it is an incredibly computational intesive task to model protein folding and even then its not perfect because it is a problem that is highly susceptible to variability based on changes to the initial conditions.

Anonymous

Anonymous Thu 17 Oct 2019 18:38:51 No.11067684 Report

Quoted By: >>11067711 >>11067737

>>11067655
so proteins interact toghter ''folding'' into new proteins?
Why can this be used against cancer?

Anonymous

Anonymous Thu 17 Oct 2019 18:49:33 No.11067711 Report

Quoted By: >>11071089

>>11067684
Not the other guy, but proteins literally do everything in the body. You're question is like asking "why would car mechanics be interesting in knowing the properties of car parts". We want to know how proteins fold, because if we can't, we cant model how the body works.

Anonymous

Anonymous Thu 17 Oct 2019 19:02:03 No.11067737 Report

Quoted By:

>>11067684
Not sure if trolling. But they are made up of subunits called amino acids that form long chains called poly peptides. There are are 20ish different amino acids with differnt chemical properties and depending on the order and the environment they will fold up into different 3d structures. Multiple Protiens can come together to form complexes and quartinary structures but you dont typically think of it as them folding into new proteins. In general they will have either structural funtion (like holding shit together etc) or enzymatic function (making chemical reactions happen faster). The function of them is pretty much determined by that 3d structure so being able to predict it accurately from amino acid sequence alone would be incredibly powerful and have really broad implications for pretty much everything not just cancer.

Anonymous

Anonymous Thu 17 Oct 2019 19:03:54 No.11067744 Report

Quoted By:

>>11066934
Charmm :^)

Anonymous

Anonymous Thu 17 Oct 2019 19:55:38 No.11067875 Report

Quoted By: >>11070760

>>11067120
I just want to simulate a few proteins, shouldn't be that demanding.

Anonymous

Anonymous Fri 18 Oct 2019 03:59:13 No.11069169 Report

Quoted By:

>>11066934
Molecular dynamics?

Anonymous

Anonymous Fri 18 Oct 2019 16:35:42 No.11070760 Report

Quoted By: >>11071072

>>11067875
Perhaps not for coarse grained stuff, but that's not interesting at all.
You definitely need a supercomputer because all the different interactions, not only intramolecular, but also with the solvent and with the binding partner (if you're studying protein folding, we're talking about self interaction, or potentially with a chaperone or a scaffold).
The gist is this: energy levels between atoms are calculated at every timestep, so just to model a few femtoseconds of protein interaction you'll need significant amount of computing power.

Anonymous

Anonymous Fri 18 Oct 2019 16:48:19 No.11070780 Report

Quoted By: >>11071079

>>11066934
>Looking for something that can resolve the structure from aa sequence and simulate protein - protein and protein - substrate interactions.
That’s 3 separate projects that will all take considerable time. If all you have is an AA sequence, you’ll have to find homologs and try to build the structure, which requires a supercomputer like another anon said. And the issue is that this won’t be perfect, so potential binding pockets might not be predicted properly preventing you from simulating protein and substrate interactions. Your best best is to identify the binding region of the AA seq. protein and only predict the 3D coords of that specific part since the rest probably won’t play any role.

Anonymous

Anonymous Fri 18 Oct 2019 18:16:34 No.11071041 Report

Quoted By:

>>11066934
there are some MC/MD python programs around, shouldnt be hard to find

Anonymous

Anonymous Fri 18 Oct 2019 18:33:20 No.11071072 Report

Quoted By:

>>11070760
Not really, if the amino acid sequence isint large he could just gromacs that shit for a week or two on a cheap VPS for a few nanosecond simulation. However i doubt that it would give him anything valuable. And you don't need significant computing power for a femtosecond timesteps if you are using forcefield methods with OP will most likely use.
>energy levels
When did people start running electronic structure calculations on whole proteins

Anonymous

Anonymous Fri 18 Oct 2019 18:35:42 No.11071079 Report

Quoted By: >>11071211

>>11070780

Is resolving the protein structure really that intensive? Obviously, it's a complex problem, I get it, but is it completely unfeasible to do that on a home computer? Would it take days or even more? From what I've read one way they do it is by some random process, but that seems very inefficient.

Maybe I'll just stick to proteins with known structures then. I don't really need to start from aa sequence, just wanted to do everything from scratch. I guess protein - protein is much less computationally intensive?

Anonymous

Anonymous Fri 18 Oct 2019 18:39:04 No.11071089 Report

Quoted By:

>>11067711
Not just the body, but every living organism on earth. Not only that, but if we can create custom proteins, we have effectively figured out how to create nanobots. Nanotechnology is cool and all, but we are no where close to what life has evolved to do with proteins, and presumably it will stay that way for quite some time.

Anonymous

Anonymous Fri 18 Oct 2019 19:23:16 No.11071211 Report

Quoted By:

>>11071079
>Is resolving the protein structure really that intensive?
Yes.

>Obviously, it's a complex problem, I get it, but is it completely unfeasible to do that on a home computer?
You could definitely do it on a home computer, it will just take a few weeks or months to finish.

>From what I've read one way they do it is by some random process, but that seems very inefficient.
This is a problem for a very long time, the algorithms used now are the best we have so far.

>Maybe I'll just stick to proteins with known structures then. I don't really need to start from aa sequence, just wanted to do everything from scratch.
To find protein structures use the Protein Data Bank website (rcsb.org). This website contains crystallized proteins that were X-rayed to put it simply. Remember, the lower the resolution, the better quality and more accurate the computational protein structure is to real life. Aim for structures ~2.0A (more or less) and below.

>I guess protein - protein is much less computationally intensive?
lol, depending on the size of the proteins it will still take some time. I do all drug docking on supercomputers, and that's between small molecules and a protein. I've never done protein-protein docking before, but it's definitely more intensive that protein-ligand obviously. How large are the proteins you're using by the way? For what it's worth, there are webservers available that perform docking, you'll have to search them up but prepare to wait a while in queue.

Capcode	All Only User Posts Only Moderator Posts Only Admin Posts Only Developer Posts
Show Posts	All Only With Images Only Without Images
Deleted Posts	All Only Deleted Posts Only Non-Deleted Posts
Ghost Posts	All Only Ghost Posts Only Non-Ghost Posts
Post Type	All Only Sticky Threads Only Opening Posts Only Reply Posts
Results	All Grouped By Threads
Order	Latest Posts First Oldest Posts First

Your latest searches