Abstract: ObjectiveIn order to provide reference basis for improving expression level of exogenous genes in banana and breeding disease-resistant banana,codon composition and usage bias of banana genome were analyzed,and the factors affecting codon usage bias were studied.MethodA total of 36242 high-confidence protein sequences in banana genome were used as data source.Codon composition and usage parameters were analyzed to determine optimal codons using CodonW 1.4.4,then correlation between usage parameters were analyzed.ResultCDS sequences of 36242 high confidence protein coding genes were selected from Banana Genome Database,their average length was 1035 bp.GC content of these genes ranged from 3.0% to 75.8%,with the average of 50.4% for whole genome.But there were only 13 sequences with GC content being less than 20.0%.Occurrence frequency of G or C in the third position of synonymous codon was 52.9%,which was higher than that of A or T.The effective number of codon(ENC) of banana genome ranged from 20.0 to 61.0,with the average of 50.7.A total of 17 optimal codons were found from banana genome,15 of which the third position of synonymous codon was G or C.Furthermore,length of gene coding regions was positively correlated with ENC.In other words,the longer gene coding region was,the weaker usage bias of codons ending with G or C was.And genes with length of 400-600 bp showed higher expression level.ConclusionThe codon usage bias for most of genes is weak in banana genome,but a few of genes show stronger codon usage bias,and prefer codons ending with G or C.Meanwhile,the codon usage bias of banana genome is affected by some factors like composition of nucleotides,mutations,natural selection etc.