p-values (exact/pseudo, one/two-sided) and consistency with SparCC

[Charlie-Hong-smu]

Hello,

Thank you for the excellent fastspar tool. For our methods section, I have three quick questions about fastspar_pvalues:

1. One-sided vs. Two-sided: Are the calculated p-values one-sided or two-sided by default? Is this controllable?
2. Exact vs. Pseudo:

• Could you briefly explain the statistical approach for the default "exact" p-values?

• Which method is generally recommended for publication: the default exact or the --pseudo p-values?

3. Consistency with original SparCC: If I use the --pseudo option in fastspar to match the permutation approach of the original SparCC, should I expect the resulting correlation and p-value matrices to be identical? Or are there known minor algorithmic differences?

Thank you for your time and for this great tool.