《学术论文集锦[开错]》紫微客

6、006

　　4.Methodology and Basic Result Analysis

　　4.1 Data preprocessing and Classical statistical analysis

　　In the tumor chemosensitive assay in vitro of this study, the concentration was set at 5 points and 8 points, and the control group and the experimental groups were both repeated for three times, so the original data obtained from the experiment is the fluorescence intensity(SI). Then, in the process of data preprocessing, we got the cell inhibition rate according to the following formula: inhibition rate = (control SI- dosing SI value) / control SI*100%, and then averaged for three replicates. Cell inhibition rate is the reflection of the degree of cell response to drug in dose-response experiment, which is the basic data set of this project.
　　After cell inhibition rates were obtained, it is first necessary to conduct the inter-group difference analysis in order to compare whether there are significant differences between disease groups, between normal groups, between disease and normal groups, and between different dose groups. Firstly, I used paired T-test (parametric test) or Wilcoxon rank sum test (non-parametric test) for comparison of differences between two groups, and One-Way ANOVA (parametric test) or Kruskal-Wallis test (non-parametric test) for comparison of differences between multiple groups. It is worth mentioning that the premise of parameter test is that the population of each group of samples all obey normal distribution and meet the homogeneity of variance. Therefore, before comparison of differences between groups, normality test and homogeneity test of variance must be conducted on sample data. Here, Shapiro-Wilk test is used for normality test and Levene test for homogeneity of variance test.

　　In addition, ANOVA is a statistical method used to test whether the mean values of multiple populations are equal, which can only show that the differences between groups are statistically significant. Therefore, when the test results show significant differences, Post-hoc multiple comparison are needed to analyze the differences between the two groups. There are many statistical methods that can be used for post-hoc multiple comparison. For ANOVA, the Least Significance Difference (LSD) method is used in this study, while for Kruskal-Wallis test, Nemenyi test is used for post-hoc multiple comparison. Finally, the test standard of the above statistical methods is to accept the null hypothesis when p-value is greater than 0.05, and all of them are carried out in R.