BioHEL is an evolutionary learning system designed to handle with large-scale bioinformatic datasets. BioHEL is strongly influenced by the GAssist Pittsburgh LCS, inheriting from it some main mechanisms. However, the main learning paradigm differs from the LCS standards to make this system more suitable for large scale domains. Moreover, a novel meta-representation called AKLR and a CUDA-based evaluation process are used to speed up the evaluation process, making possible for this system to solve very large and complex real life problems in less time.