Global energy minimization by rotational energy embedding.

Area of Science:

• Computational chemistry
• Molecular modeling
• Biophysics

Background:

• Predicting preferred molecular conformations is computationally intensive for large molecules due to vast numbers of local energy minima.
• Energy embedding methods offer a promising strategy for identifying high-quality local minima, though not always the global minimum.

Purpose of the Study:

• To introduce a novel variation of energy embedding, termed rotational energy embedding, for more efficient conformational prediction.
• To demonstrate the efficacy of rotational energy embedding in locating near-native conformations for peptides.

Main Methods:

• Developed rotational energy embedding, a method that projects high-dimensional molecular representations into three dimensions via internal rotations.
• Generalized multidimensional internal rotations to mimic torsion angle variations in three dimensions.
• Applied the method to avian pancreatic polypeptide and apamin using only their amino acid sequences and a potential function.

Main Results:

• Rotational energy embedding successfully identified conformations very close to the native structures for avian pancreatic polypeptide and apamin.
• The new method overcomes limitations encountered in ordinary energy embedding techniques.
• Accurate conformational prediction was achieved without prior structural information.

Conclusions:

• Rotational energy embedding is an effective computational technique for predicting molecular conformations.
• This method significantly advances the ability to determine peptide structures from sequence data.
• The approach holds potential for broader applications in molecular modeling and drug discovery.