The most intuitive way to evaluate the quality of a model in Rosetta is to use Rosetta's scoring system, which is often referred to as an energy function. We use some scoring items directly related to energy to score the structural coordinates of the protein.
Rosetta energy item type and classification
The Rosetta energy function is a functional form obtained by weighting a series of measurable geometric statistics or classical physics interaction energy. Under the condition of given atomic coordinates, evaluate the magnitude of the interaction energy between atoms. In the previous version of Rosetta, the unit of energy is REU (Rosetta Energy Unit), and the latest version of the full-atom scoring function (ref2015) has been corrected, and the current unit is kcals/mol.
In terms of interaction type, Rosetta's scoring items are usually divided into three categories:
One Body: Usually this type of scoring item is only related to the conformation of a single amino acid, such as the dihedral angle of the skeleton, the rotamer conformation of the side chain, etc.
Two Body: This type of scoring item is related to two amino acids, such as van der Waals interaction and electrostatic interaction
Whole Body: Consider the energy of the protein from the overall geometric properties or other indicators, such as the protein's gyration radius, secondary structure composition and other statistical quantities.
It can be distinguished from the fitting method of scoring items, which can be divided into physical potential energy items and statistical potential energy items:
- The physical potential energy term is usually calculated from the classical formula of molecular interaction defined in physics, such as the LJ potential function of van der Waals force and the electrostatic potential function of Coulomb force.
- Statistical potential energy items are generally obtained from the protein structure database. There is a more important assumption here. In the data set, the higher the probability of occurrence or conformation, the lower the energy, and the rarer the higher the energy (Boltzmann distribution). Therefore, it can be directly taken by measuring the frequency. The magnitude of the logarithmic approximation energy. For example, by counting the distribution of the phi and psi angles on the protein backbone, the Ramachandran diagram can be statistically obtained. The more enriched points, the lower the energy of these backbone conformations, which are frequently observed in nature.
After getting multiple scoring items, usually given some data set conditions, by fitting weights, a relatively good scoring function can be performed, and the same is true for Rosetta energy function fitting. For example, through the fragment assembly of ab-inito, a large number of predicted protein structure models are obtained. By trying to adjust and combine the scoring weight of each item, the energy of the model closest to the crystal is tried to fit the lowest energy state.
Specific explanation of Rosetta energy term
1. LJ potential
The function describing the interaction of van der Waals force is related to the distance between two atoms. The parameter comes from the CHARMM force field.
2. Electrostatic potential
The distance-dependent electrostatic potential function is derived and optimized from CHARMM. The value of this term is related to the number of charges of the two atoms, the dielectric constant of the environment, and the distance between the atoms.
3 Hydrogen bond interaction
Rosetta uses the hbond term together with fa_elec to calculate the energy of the hydrogen bond.
The hbond term is derived from the Top8000 high-precision structure database. First, the polar interaction pairs inside the protein are separated, and their dihedral angle, bond angle, distance between donor and acceptor atoms are fitted. The energy distribution.
4. Disulfide bond energy (dslf_fa13)
The Rosetta disulfide bond item is also similar to hbond, derived from the database, and is related to the conformation of the specific disulfide bond.
5. Side chain conformation energy
This item is used to evaluate the energy value of the side chain conformation. The calculation principle is to query the probability of occurrence of the Rotamer type from the dunbrack rotamer database, and estimate the energy value by the deviation from the average conformation of the rotamer.
* For Research Use Only.
Related Services
