Studies On The Relationship Of Methanobacteriaceae Thermostability And I Protein Sequence And Structure By Bioinformatics Method

Thermophile protein is thermostabilization protein from thermophile microorganism. It can keep active for a long in high temperature. Through comparing of exthermophilic, thermophilic and normal archaebacteria,we find that many proteins have similar structure and function, but they have distinct thermostability. To know more about the reason of thermostability, more about protein folding and structure and function and evolution and the meaning of designing of preotein molecule must be known. So the study select three Methanobacteriaceae to analyse their sequence and structure and find cause of thermostability.Genbank and COG and PDB and PIR and PSD and other databases are used. The sequences and structures of Methanobacteriaceae are analysed using Matlab to find the reason of thermostability. And BP neural net and Support Vector Machine are used to forecast thermostability of Metnanobacteriaceae protein.First, the gene sequence of three kinds of Methanobacteriaceae including codon bias and CG content are studied. Gene include 20 kinds amino acid and information of influencing protein thermostability. Finding these informations can help improve veracity of forecasting protein structure. Through studing the gene sequence, we find codon bias and codon miss rate are positive correlation. They are positive with thermostability.Then I count protein primary structure sequence include sequence length and amino acid content. Studes find that the sequences of thermophile protein are long and have much polar residue. Thermophile and unthermophile protein sequence are selccted from protein primary structure and their eigenvector are distilled to train BP neural net. The net is uesed to forecast thermostability of protein primary structuer. The result is perfect, local prediction rates of three Methanobacteriaceae are 86.7%, 63.4% and 93.3%.Downloading information of protein secondary structure from PDB, Through counting 8 kinds of structure character fact is finded thermophile protein have more Alpha helix and less Beta sheet. Eigenvetor is distilled from protein secondary structure to train Support Vector Machine to forecast thermostability. The result is discriminative, thouth it is nor perfect than the result of protein primary structure.Thermophile and unthermophile genes are translated to protein primary structure and distill eigenvetor to forecast thermostability in BP neural net. The experiment indicates that protein'gene is thermophile and the protein is also thermophile. Ensembling several DIMLP net and network tool are used to forecast protein secondary structure using thermophile and unthermophile protein primary structure. Predicted protein secondary are counted and eigenvector is distilled to input Support Vector Machine to forecast thermostability. The counting indicated that I find that protein secondary structure predicted from thermophile protein secongdary is also thermophile, and character is in evidence.