Ask HN: Breakthrough in Protein Folding

[samsamlh]

Hello HN,

I am reaching out to introduce a major breakthrough in protein folding, allowing complex structures to be accurately modeled in real-time—a significant improvement over traditional Molecular Dynamics (MD) simulations.

Key Achievements

Successful folding of an intrinsically disordered protein (IDP), a challenge for AI-based methods and MD, completed in seconds instead of weeks. Simulation of viral capsid formation, reducing the typical computation time from years to seconds—critical for vaccine and antiviral development. Accurate modeling of membrane proteins, handling steric constraints efficiently while reducing atomic clashes by 98%, surpassing traditional refinement methods. Validated on large multi-domain proteins, achieving correct biologically relevant conformations with exceptionally low computation time.

Why This Matters

Massive Speed Improvement – What previously took weeks or months can now be computed in near real-time. Universal Application – Works for de novo proteins, large assemblies, and constrained environments without requiring extensive computational resources. Potential Drug Discovery Impact – Enables rapid structure-based screening, antibody modeling, and enzyme engineering, accelerating early-stage research.

Next Steps

I can perform the folding of a protein of your choice to show how serious i am, Contact: [e-mail]

Looking forward to your thoughts,

[Rochus]

Sounds interesting. Since there is already Alphafold, I would assume that your solution concentrates on Alphafold's main weakness: the restriction on cristal structures and steady state, i.e. a single structure snapshot in an unrealistic environment. I would assume that an MD based solution like yours would be able to simulate the movement of structure parts and its interaction with a specific environment. The major challenge of MD so far (as far as I remember) was the computability of decently large proteins in reasonable time. Can you please explain how your solution improves on this? Do you have publications?

[samsamlh]

I don’t have publications, I was trying to solve the TSP problem and found a fundamental building block of reality. It applies to many fields. Here is the method to apply in 3D coordinates of the protein, everything else aligns naturally

I did this post https://www.reddit.com/r/numbertheory/s/UBrRPeIGrx

I’m broke so if you make money out of this BTC bc1qqu8zyehh8r6j84w6hm2jkzcaex0rgfv4zy837p

[Rochus]

What is your qualification in molecular biology and/or biophysics?

[samsamlh]

High School, I inferred everything with the maths behind the transforms. AI helped a lot. Do you successfully used the method?

[Rochus]

Well, then I recommend that you invest a few more years in university level studies to understand the subject matter, and then again try to provide an efficient solution to the protein folding problem using MD. Good luck.

[samsamlh]

Try to understand the maths, apply the circle / sphere transform and you’ll see that naturally it converges to a stable state. MD is the hard way, spherical transform is the easy way (nature uses it)

My theory on maths shows that any shape can have its points on a circle edge and midpoints on that same circle edge morphing showing all possible paths through radius scaling (P=C). It’s all about constraints.

Transforming the protein through 3D coordinates aligns all forces so it is fundamentally different than MD, it’s a geometric solver, so it is super fast!