| Phylogenetic inference is an important branch of computational biology.Its main purpose is to reconstruct the evolutionary history and relationships between species,understand when and what species formation events may occur,and study the evolution of species.In order to solve the problem of the instability of phylogenetic tree construction caused by missing data in the morphological data,firstly,we combine the character description and prior knowledge of morphological data to generate the character hierarchical relationship and use the hierarchical relationship between characters to impute the missing value.Perform estimation to obtain a complete data set;then,the inapplicable Fitch algorithm is used to distinguish and measure the state transition between applicable data and inapplicable data;finally,combine the principle of maximum parsimony and maximum likelihood for multi-object phylogenetic tree construction,generating a Pareto optimal phylogenetic tree,avoiding the limitations of building trees under a single principle Compared with the existing phylogenetic analysis methods,it can well solve the problem of phylogenetic tree construction instability caused by missing data and unsuitable data.It also provides a collection of trees under multiple tree building principles.It provides a basis for biologists to study the evolution of species.The main work of this paper is as follows:1)Construct and formalize the hierarchical relationship among the characters according to the logical association among the morphological characters.Combining characters hierarchical relationship information,we proposed an algorithm for estimating missing values ??of morphological data based on feature hierarchical relationships.The experimental results show that the missing value estimation algorithm based on character hierarchical relationship is better than common missing value estimation methods in both accuracy and topological difference rate.The phylogenetic tree constructed by the imputed data set are more similar to the standard phylogenetic trees,which solves the problem of the accuracy of phylogenetic trees under the influence of missing data.2)For the processing of inapplicable data,compared with common methods of processing inapplicable data,such as treating inapplicable data as missing data or extra states,the inapplicable Fitch algorithm can better distinguish and measure applicable data and inapplicable data.It uses to constructing of multi-objective phylogenetic trees to better solve the problems caused by inapplicable data.3)In order to avoid the possibility of conflicts in the same data set under multiple tree building principles,this paper combines the principle of maximum parsimony and maximum likelihood to propose a multi-objective phylogenetic tree construction algorithm,which can satisfy multiple tree building principles.Experiments show that 65.4% of the 500 tree structure constructed by the method proposed in this paper is consistent with the phylogenetic tree generated by existing software.In terms of topological accuracy,the topological similarity of the proposed method(only imputation)is 82.2% higher than maximum parsimony,maximum likelihood,Bayesian inference,and neighbor-joining method.When the morphological data contains missing,the average topology similarity of the proposed method(firstly imputation and then building trees)is83.26%.Therefore,compared with the existing methods,the proposed method has a good application prospect. |
