There are three main methods for protein structure prediction: homology modeling, folding recognition and de novo calculation. All those methods are used to predict unknown protein structures. Since the amino acid changes generally occur in the inflection region on the surface of the protein, the main chain structure of the protein, especially the hydrophobic core, is not susceptible to sequence changes. Therefore, the homology modeling method is the most widely used. Homology modeling uses the protein crystal structure homologous to the target sequence as a template to predict the three-dimensional structure of the target sequence. The conservation of the tertiary structure of a protein is far greater than that of its primary structure. Theoretically, secondary or tertiary structure can be obtained from the primary structure, that is, the amino acid sequence can determine the structure of a protein. Therefore, homology modeling is a practical method for predicting structure from sequences. This method has two prerequisites: First, there are one or several resolved structures in the homologous protein of the target sequence. Second, the homology between the target sequence and the protein is high.
The definition of homology modeling
Homology modeling is also called comparative modeling. Its principle is that similar sequences have similar structures, that is, two sequences with homologous relationships have similar structures. When the homology of the two sequences is greater than 30%, the homology of the sequences can imply that the two are structurally similar. The higher the homology of the sequences, the higher the accuracy of the structural model.
The basic principles of homology modeling
1. The structure of a protein is uniquely determined by its amino acid sequence, by the primary structure. In theory, it is enough to obtain its secondary and tertiary structure.
2. The conservative type of tertiary structure is much larger than that of primary structure. And the sequence identity of template protein and target protein needs to be greater than 30%
The main steps of homology modeling
It generally includes the following seven steps:
1. Select template and alignment sequence: search for template protein with known crystal structure from protein database and use BLAST and other tools to align the sequence;
2. Determine whether there is a template available: template sequence and target sequence similarity "30%;
3. Construct the main chain structure: apply the atomic coordinates of the template structure to the target to generate a basic main chain skeleton and then adjust the main chain atom position to make the bone structure conform to the stereo-chemical principles;
4. Construction of ring chain: One method is to model based on the known ring zone structure. Another method is to predict from scratch based on the principles of quantum chemistry;
5. Modeling and optimization of the side chain: use the target sequence fragments to search the rotamer database to obtain similar fragments, and then construct the side chain structure of this fragment according to its spatial orientation. Finally, the energy is minimized to find the lowest energy point, that is, the stable conformation;
6. The overall structure is optimized: there is usually unreasonable contact between atoms in the three-dimensional model obtained through the above process, which needs to be eliminated by methods such as simulated annealing and molecular dynamics.
7. Structural evaluation: The most common evaluation criterion is RMSD, which represents the root mean square deviation of the corresponding atoms between the target protein and the template protein. It can also be submitted to the SAVES server for verification.
With the improvement of prediction methods, homology modeling has been widely used in drug design and protein design. It can interpret the relationship between protein structure, function and sequence. In the post-genomic era, homology modeling will surely become a bridge between genetic engineering and protein engineering.
With comprehensive modeling services, designing and engineering novel antibodies with desired therapeutic properties, Creative Biolabs offers a variety of antibody structure modeling services for researchers: