Computational Simulation For The Sequences, Structures, And Activities Of Biological Macromolecules

With the completion of Human Genome Project and the rapid development of computer technology, bioinformatics has become a very popular research area. The contents of bioinformatics are very rich. It mainly regards the information on DNA sequences as the source, and performs simulation and prediction of the spatial structures of proteins after obtaining the information about the coding regions of proteins. At last it executes necessary drug design according to the function of special proteins. This dissertation mainly focuses on the sequences and structures of biological macromolecules. The contents of this dissertation are on sequence similarity analysis, structural classification, the relationship between structures and activities, and drug design. Novel sequence comparison and structural classification methods are proposed in this dissertation to overcome the limitations of traditional comparison methods.In order to reveal the relationship between structures and activities of biological molecules, we study the relationships between structures and activities of cyclic peptides, Dichotomins A and B, by means of quantum chemistry method. Molecular dynamics simulation based on molecular field is proposed to study the conformational change of cyclic peptide, Stylopeptide 1, in different solvents, and the interactive mechanism of HIV-1 protease with inhibitors.The major contents of this dissertation are as follows:Firstly, aiming at some shortcomings such as degradation phenomena, computational complexity, and the loss of information among nucleotides, accompanied by existing graphical representation and invariant methods, a new graphical representation (PNN curve) and invariant (y-M) method based on DNA neighboring nucleotides is given. The coding sequences of the first exon of P-globin gene of eleven species are analyzed by this method, we find the distribution differences of PNNs are helpful for the study of the phylogenetic relationships of the eleven species. Both Gallus and Opossum are very dissimilar to the others, while Human, Gorilla and Chimpanzee are very similar. These results are consistent with the phylogenetic fact. Comparing to traditional graphical representation, PNN curve has two major advantages:(1) the curve does not overlap or intersect with itself and considers the information of neighboring bases, so it avoids loss of biological information; (2) It is more sensitive to the change of nucleotides. In addition, three invariant methods, the three-component vector comprised of the gravity coordinates of cyclic peptides, the normalized leading eigenvalues of E matrix and L/L matrix, are used to classify eight cyclic peptides whose conformational characters are known. By comparison, we find the normalized leading eigenvalues of L/L matrix is the best method to classify the conformations of cyclic peptides.Secondly, quantum chemistry method is used to optimize the initial structures of Dichotomins A and B, and calculate the 13C chemical shifts of the optimized structures of Dichotomin B. The relationships between structures and activities of Dichotomins A and B are studied. The computational results show that intramolecular hydrogen bonds and P-turns may be major factors of maintaining the conformations of cyclic peptides A and B. Comparing the structures of the two cyclic peptides, we find they both have two intramolecular hydrogen bonds. But, in the optimized geometries of A twoβ-turns are formed, while in the optimized geometries of B only oneβ-turn exists. The structural differences may be the main reason of leading to their different activities.Thirdly, molecular dynamics simulation method is used to study the solvent (water and methanol) effect on the conformation and activity of cyclic peptide, Stylopeptide 1. The simulation results show that some hydrogen bonds andβ-turns exist all the time in methanol solvent. Whereas no hydrogen bonds or turns always exist in water. Solvent effect on the conformation of Stylopeptide 1 may be the reason that this cyclic peptide exhibits different activities.Lastly, molecular dynamics simulation method is used to study the interactive mechanisms of CRF01_AE human immunodeficiency virus type 1 protease (HIV-1 PR) and mutant AE HIV-1 PR with inhibitors LPV and NFV. The computational simulation results show that V82F mutation make the conformation of 79's loop region change, which lead to the rotation of the side chains of residues D25 and 150'and the asymmetrical deformation of the conformation of the binding cavity, and some atoms of LPV are dislocated. So some interactions between protease and LPV are weakened, and the binding affinities of LPV with protease are reduced. The existences of the dodecahydroisoquinoline ring, the 2,6-methylbenzamide, and the N2 atom in the protease inhibitors are favorable to the binding of inhibitors with protease.