The history of computing the correct identification of eukaryotic gene organization is an issue that long. Despite the fundamental importance of note that the precision of the genes encoded in the genome sequenced new, the accuracy of the structural genes predicted not been evaluated by critics, mainly due to the lack of a proper assessment. We present results structural gene realize how many genes arranged in order to predict the amino acid sequence translated from the same gene from the genome of many. How to provide rich information about the reliability of each predicted gene structures. We also have developed a method that emphasizes that efforts to improve the structure of the gene defect prediction algorithms based on the consensus sequence alignment married or homologs reliable as a template. The application of our approach to cytochrome P450 and ribosomal proteins from 47 genomes of plants indicated that 50 ~ 60% of the annotated genes are likely to have some flaws. While more than half of the defective genes that may be destroyed from within, they are fake or parts of genes in a sequence unfinished or non-compliance with isoform production. defects found in the majority of candidate genes to be remedied. By the way, our tune over and over again. Our method is applicable to the families of the gene encoding the protein domain structure their stable evolution. It is also possible to apply our method to a family of genes from all kingdoms of life, not just plants.