[en] To understand codon usage features of ICE1 gene in radish, the sequence of ICE1 was analyzed via programs of CUSP and Codon W, and it was also compared with other 10 plant species and genomes. The results showed that ICE1 in radish was bias towards the synonymous codons with A and T at third codon position. According to the RSCU (Relative Synonymous Codon Usage), 12 codons were high-frequency. Compared with the codon usages of other radish genes, bias towards of 19 codons were the same. There were some differences in codon bias of ICE1 among different species. The cluster tree of ICE1 based on CDS revealed a better evolutionary relationship among different species than the result analyzed by RSCU. The differences of codon usage between ICE1 and rice genome was larger than that between ICE1 and some cruciferae plant genomes (Arabidopsis, cabbage). This means something should be done to ICE1 to heterologous expression of ICE1. This study not only preliminarily reveals the codon usage pattern of ICE1 of radish, but also provides a certain theoretical basis to further molecular genetic study. (authors)

[en] To establish and validate a radiomics model based on multiparametric magnetic resonance imaging (MRI), and to predict microsatellite instability (MSI) status in rectal cancer patients. A total of 199 patients with pathologically confirmed rectal cancer were included. The MSI status was confirmed by immunohistochemistry (IHC) staining. Clinical factors and laboratory data associated with MSI status were analyzed. The imaging data of 100 patients from one of the hospitals were used as the training set. The remaining 99 patients from the other two hospitals were used as the external validation set. The regions of interest (ROIs) were delineated from T1-weighted imaging (T1WI), T2-weighted imaging (T2WI), diffusion-weighted imaging (DWI), and contrast-enhanced T1WI (CE-T1WI) sequence to extract the radiomics features. The Tree-based approach was used for feature selection. The models were constructed based on the four single sequences and a combination of the four sequences using the random forest (RF) algorithm. The external validation set was used to verify the generalization ability of each model. The receiver operating characteristic (ROC) curves and the area under the curve (AUC) were plotted to evaluate and compare the predictive performance of each model. In the four single-series models, the CE-T1WI model performed the best. The AUCs of the T1WI, T2WI, DWI, and CE-T1WI prediction models in the training set were 0.74, 0.71, 0.71, and 0.78, respectively, while in the external validation set, the corresponding AUCs were 0.67, 0.66, 0.70, and 0.77. The prediction and generalization performance of the combined model of multi-sequences was comparable to that of the CE-T1WI model and it was better than that of the remaining three single-series models, with AUC values of 0.78 and 0.78 in the training and validation sets, respectively. The established radiomics models based on CE-T1WI or multiparametric MRI have similar predictive performance. They have the potential to predict MSI status in rectal cancer patients. A radiomics model for the prediction of MSI status in patients with rectal cancer was established and validated using external validation. The models based on CE-T1WI or multiparametric MRI have better predictive performance than those based on single unenhanced sequence images. The radiomics model has the potential to suggest MSI status in rectal cancer patients; however, it is not yet a substitute for histological confirmation.