Evolutionary Pattern Analysis Of Retro-Chimeric Genes In Rice

Theodosius Dobzhansky once said "nothing in biology makes sense except in the light of evolution". With the continuous development of molecular biology, particularly in genome sequencing mean, evolutionary biology received a more extensive research space. Molecular evolution is one of the fastest growing fields in evolutionary biology analysis accompany with huge accumulation of genomic sequence data. Molecular evolution research including the analysis of the various sequence polymorphisms, such as restriction polymorphism, protein polymorphism, single nucleotide polymorphism analysis, according to these information sites, people can build phylogenies (species tree or gene tree), can estimate the evolutionary rates, can sort out the evolutionary pattern, can calculate the birth time of new genes, and can predict gene function and its adaptation to the environment. As the genomic sequence data of various species, the genomic sequence data of close related species, the genomic sequence date of varieties and even the genomic sequence data of individuals increasing quickly, using genomic sequence polymorphisms to study the molecular evolution is becoming increasingly popular, this situation make a higher demands to the data analysis methods.In this research, we provide a formula to calculate the number of all the possible branch models for a specific phylogeny with n branches, and the calculation results show that it is a gigantic number consistent with the Bell Number. Since the huge number of all possible branch models, it is very difficult to make a global calculation. We present a dynamic program method to assistant CODEML to explore the specific branch model which has the maximum likelihood in variety substitution rates, then it is suggested using likelihood ratio test (LRT) or Akaike information criterion (AIC) to judge the better model to get the final optimal model, we call this whole method as OBSM (optimal branch specific model) method.There're three methods included in OBSM method, the method 1 cost fewest time to explore the optimal model while method 3 seems to catch the best optimal model. Using these OBSM methods, we calculate all the possible model of the lysozyme data and get the final optimal model. The comparison among final optimal model, final best model from global calculation and the conventional hypothesis model, show that the log likelihood value of optimal model is very close to the best model and it is significant better than conventional hypothesis model which is based on the priori knowledge. Then, we analyzed 50 data sets from 40 papers which are randomly pick up, the results show the same conclusion that in 47 data sets, the final optimal model suggested by OBSM method is significant better than hypothesis models while in other 3 data sets, there're no difference. Further more, in the 17th case, the substitution rate evaluated by the final optimal model suggest that the branch Sinocyclocheilus purpureus is under positive selection. This result shows that the conventional hypothesis model obviously misses some valuable information. Through our efforts, OBSM method mended the pre-existing deficiencies in conventiaonal hypothesis model method. It helps the user get a reasonable optimal branch model in an acceptable time and avoid the huge time cost in global calculation; it also helps the user avoid the limitations or the subjectivities when using the priori knowledge to pick up the hypothesis models. We constructed a web service called OBSM (optimal branch specific model, http://obsm.ncpgr.cn) and also build some script programs which can run on the different OS, available for the public users.The mechanism of the birth of new gene and the evolutionary pattern of new gene is another important field in molecular evolution. Here, the new gene we talk about is the genes that formed by insert the cDNA into a new genomic position which is reverse-transcribed from the mRNA and then recruit nearby sequences to form a new structured gene; we called it retro-chimeric gene. Because of the processing processes of mRNAs, the newly duplicated homologous genes always lack introns, have a poly-A tail and have short flanking repeats. Since professor Long identify the first retro-chimeric Jingwei gene in 1993, more and more new genes have been found. The rich dataset of retro-chimeric genes in rice provided an opportunity to investigate the evolutionary patterns of the retro-chimeric gene pairs. In this research, using OBSM method, we analyze 24 pairs of new genes and their parental genes in rice, the substitution rate of final optimal branch models show that 7 pairs of genes are under positive selection. According to the analysis, we found three evolution patterns:(1) The retro-chimeric genes were rapidly substituted in the initial stage of the new gene lineage under positive selection; (2) The parental genes evolved rapidly soon after the retro-chimeric genes were formed whereas the new genes evolved slowly in evolution (3) The parental genes evolved rapidly after some uncertain period of the retro-chimeric genes were formed whereas the new genes evolved slowly in evolution. The first point is partially consistent with the pattern revealed by Jones and Begun while the other two point, we have not seen any report.Then, molecular clock analysis results show that three new genes, RCG1, RCG2 and RCG3 are very young (3MYA) or young (7MYA,15MYA). The chip data of the entire life cycle of rice show that the RCG3 have an expression peak in Zhenshan 97 (a variety of cultivated rice) at infection period at calli. This evidence indicate that this new gene have a stress reaction against the infection, and it is in consonance with the independent evidence, that it encodes Leucine-rich proteins, and has a high similarity with the Vel gene which have been shown to be resistance against verticillium wilt disease.