In the Seurat package for scRNA-seq data analysis, there is a function called FeatureScatter, which can plot the scatter of the correlation between two features. I tried to make such a plot of two features in a dataset with the command:
FeatureScatter(object = obj, feature1 = 'Feature1', feature2 = 'Feature2')
and get the following image:
The Pearson correlation value is only 0.12. Do this mean that these two features are not very correlated? But looking at many lines formed by the points, it seems that they could be related.
Is there any standard for judging whether two variables are correlated or not in such a plot? Is the plot command not correct? Can the plot be improved and the calculated correlation value be corrected?
If FeatureScatter is not a suitable function to calculate the correlation, is there other functions that can do this?