Table 2.

Algorithm of estimating empirical probabilities from control/treatment data of metabolic network inhibition.

1.	Initialize a matrix MR with column=2 and row=number of metabolites; % MR[α, 1] stores the state value of the concentration of α (up or down) and MR[α, 2] stores the extracted probability, Pe(concentration(α, MR[α, 1]))
2.	for each metabolite α do 2.1. Cα = a set of concentration values of α observed in the control cases; 2.2. Mα=Mean(Cα),SDα=StandardDeviation(Cα); 2.3. Tα={τα}, a set of concentration values of α observed in the treated cases; 2.4. MR[α, 1] = Mα < Mean(Tα) ? up : down; % Decide the state value (up or down) of the concentration of α by the difference between Mean(Cα) and Mean(Tα) 2.5. MR[α, 2]=ρα=Mean({pnorm(τα, Mα, SDα)}); % Calculate the average of the integrals returned by pnorm function
3.	Apply matrix MR in the abductive SLP learning.