An algorithm and its implementation are described that use whole genetic or chemogenetic interaction profiles and identify the most informative subprofile which enables linking the studied substances with the specific molecular targets.
COBISS.SI-ID: 22874663
We have developed an efficient algorithm that can relate the structure of a small chemical with its corresponding phenotype in the model organism. The phenotype can be described with a vector of response variables (phenotype discription). The method is new to the field of machine learning as well, as previous subgroup discovery efforts focused on problems where response was defined only with a single variable.
COBISS.SI-ID: 7256404