How to get P-value by sclink_cor() function

[zeypku]

hello, I like the algorithm of sclink_cor() function. But I have two questions: 1.how to get p-value of sclink_cor() result? 2. what's the difference between sclink_cor() and cor(use= "pairwise.complete.obs") thanks!

[Vivianstats]

Hi,

sclink_cor currently does not provide p values of the robust correlation measures, but you consider running a permutation test to get the p values.

The main difference is that scLink uses a mixture model to determine which gene expression values are accurately measured and used to calculate the correlation, while cor(use= "pairwise.complete.obs") use all non-NA values.

[zeypku]

Thanks！ I found if I use NA replace the zero counts in the expression matrix and use cor(use= "pairwise.complete.obs") to calculate the correlation. Most results are the same as the scLink. But some are NA whereas scLink gives an exact number.

[ismailelshimy]

Could you kindly provide a helper script to calculate these P-values using the bootstrapping approach you kindly highlighted in the supplementary section of your manuscript? @Vivianstats

Thank you very much !

[Vivianstats]

Hi @ismailelshimy,

The permutation approach is described on page 2 of the supplementary methods (https://ars.els-cdn.com/content/image/1-s2.0-S1672022921001455-mmc1.pdf) I'd be happy to take additional questions if anything in this procedure is not clear.