FAQs

Frequently asked question

Analytical issues

1. How are the integrated variables for CCA analysis understood?

In order to grasp the correlation relationship between the two sets of indicators overall, representative two comprehensive variables U1 and V1 (linear combination of each variable in the two sets, respectively) were extracted in two sets of variables, and the correlation relationship between these two comprehensive variables was utilized to reflect the overall correlation between the two sets of indicators.

2. Random forest analysis significance scores of differential species versus differential metabolites there were two, the results of meandecreaseaccuracy and meandecreasegini were not the same, and how was the result selected?

The mean decrease accuracy plot represents how much accuracy the model has lost by excluding each variable. The more accuracy is affected, the more important the variable is for successful classification and the variables are in descending order of importance. Mean decrease in mean decrease Gini, the Gini coefficient, is a measure of how each variable contributes to the homogeneity of nodes and leaves in the generated random forest. The higher the value of the mean descending accuracy or mean descending Gini score, the more important the variable is in the model. The mean decrease accuracy is more reliable than the mean decrease Gini index if one is to distinguish which is more reliable individually, the measured trait scales and / or the number of categories differ.

3. Why is the area under the ROC curve equal to 1 for both single - and pooled omics random forest and plsda?

Because here single and pooled omics data ROC analysis was used with all the difference results, in the case of the small number of samples in each group, and the model comparison was easy to distinguish the comparison group and the control group, so the single - and pooled omics data modeling can both have a good effect.

4. Why not draw the figure without the metabolite name?

Because metabolite names are often longer, having a metabolite name results in a picture that is malformed.

5. Why do joint analyses require samples for one-to-one correspondence?

For better mining of intrinsic connections among omics.