Algorithm LOPC
1:	Zero-th order partial correlation:
2:	for each pair (xi, xj) do
3:	Calculate an estimate of the zero-th order partial correlation coefficient rij;
4:	Construct the test statistic for rij and compute the corresponding p-value p(rij);
5:	Compute the multiple testing adjusted p-value for the zeroth order partial correlation coefficient p̄(xij) across all pairs.
6:	end for
7:	First order partial correlation:
8:	for each pair (xi, xj) do
9:	Calculate estimates of the first order partial correlation coefficients rij·k for all possible xk ∈ X/{xi, xj};
10:	Select the maximum in terms of absolute value as r̂ij·k;
11:	Construct test statistics for r̂ij·k using Fisher’s z transformation and compute corresponding p-value p(r̂ij·k);
12:	Compute the multiple test adjusted p-values for the first order partial correlation coefficient p̄(r̂ij·k) across all pairs.
13:	end for
14:	Second order partial correlation:
15:	for each pair (xi, xj) do
16:	if max {p̄(rij), p̄(r̂ij·k)} < 0.05 then
17:	Proceed to compute the second order partial correlation:
18:	Calculate estimates of the second order partial correlation coefficients rij·kq for all possible xk, xq ∈ X/ {xi, xj};
19:	Select the maximum in terms of absolute value as r̂ij·kq;
20:	Compute the multiple test adjusted p-values for the second order partial correlation coefficient p̄ (r̂ij·kq) across all pairs.
21:	else
22:	Do not need to compute the second order partial correlation:
23:	Set p̄(r̂ij·kq) to be 1.
24:	end if
25:	end for
26:	Connect xi and xj iff p̄(r̂ij·kq) < 0.05.