Computers, computer program products, and methods for identifying a plurality of subjects for a clinical trial are provided. A candidate set of molecular profiles in a stored plurality of molecular profiles are identified. Each such profile has measurements for a discriminating set of cellular constituents that match the measurements of corresponding cellular constituents in a responder set of biological samples, thereby identifying the plurality of subjects for the trial from those subjects from which the candidate set of molecular profiles were derived. Each respective molecular profile in the stored plurality of profiles has measurements of a plurality of cellular constituents from a respective biological sample in a plurality of samples obtained from a first plurality of subjects. The discriminating set of cellular constituents is identified from those cellular constituents in the plurality of cellular constituents whose measurement values discriminates between the responder and nonresponder sets of biological samples.