We are trying to see significant differences between the correlation matrices of two groups using permutation test. The number of significant edges (cell of correlation matrix) are much higher than we expect with significance level of 5% (p < 0.05). The number of subjects in each group are more than 10. I just want suggestion that is it good to set significance level of 5% without caring about number of subjects in the groups or have introduce any correction factor to significance level for smaller number of subjects in the group? If yes, then what should be the maximum number of subject below which correction factor should be imposed? and what it should be?

Thanks in advance!