A method of searching one or more ligand-candidate compounds to a target biopolymer from a three-dimensional structure database, which comprises the steps of: (i) the first step of assigning hydrogen-bonding category numbers, information for calculating force-field energy, and information for generating conformations to two or more trial compounds in addition to atomic three-dimensional coordinates thereof; (ii) the second step of preparing physicochemical information about a ligand-binding region and one or more dummy atoms based on the three-dimensional atomic coordinates of the target biopolymer; (iii) the third step of estimating the most stable docking structure, wherein said step further comprises the steps of examining possible docking structures by docking a trial compound to the biopolymer while varying conformations of the trial compound, evaluating interaction energies between the target biopolymer and the trial compound, and repeating structure optimization; (iv) the fourth step of deciding whether or not the trial compound should be adopted as a ligand-candidate compound based on given criteria including the interaction energy values between the target biopolymer and the trial compound in the most stable docking structure estimated according to the step (iii); and (v) the fifth step of repeating the step (iii) and the step (iv) for all of the trial compounds.