To what extent can the 'protein folding problem' be… — Dialectica

To what extent can the 'protein folding problem' be considered solved following the success of AlphaFold2, and what specific challenges regarding protein dynamics and folding pathways remain unaddressed?

About this Question

Dialectica's answer

The "protein folding problem" is considered solved only in the specific context of predicting the static 3D native structure of highly ordered, globular proteins from their amino acid sequences Verified Answer #1, Verified Answer #2. AlphaFold2 achieved this by leveraging deep learning and evolutionary information to reach accuracy levels comparable to experimental methods like X-ray crystallography Verified Answer #3. AlphaFold 3 further expanded this capability to include holistic biomolecular complexes, such as proteins interacting with DNA, RNA, and ligands Verified Answer #4. However, because these models bypass the physical process of folding to predict a final thermodynamic endpoint, the broader biophysical components of the problem—specifically protein dynamics and kinetic pathways—remain largely unaddressed Verified Answer #1, Verified Answer #3.

Unaddressed Challenges in Dynamics and Kinetics

The temporal mechanisms and intermediate states required for a protein to reach its folded state, known as folding kinetics, are not simulated by AlphaFold models Verified Answer #1. These tools do not address Levinthal’s kinetic paradox, which questions how a protein chain finds its native conformation so quickly without sampling every possible state Verified Answer #4. Because the models rely on static snapshots from the Protein Data Bank, they struggle to map the conformational fluctuations and thermodynamic ensembles that proteins populate to execute biological functions Verified Answer #1, Verified Answer #4.

Furthermore, AlphaFold does not explicitly calculate the fundamental physical chemistry of atomic interactions, such as solvent effects or quantum mechanical forces, that drive the folding process Verified Answer #1, Verified Answer #2. While the AI identifies the global energy minimum of the native state through statistical patterns, it does not provide an understanding of the physical folding trajectory itself Verified Answer #2, Verified Answer #3. Consequently, the "physical" protein folding problem remains a major frontier for the scientific community Verified Answer #3.