Our overall approach will be to identify basic data mining operations that cut across applications and develop scalable algorithms for their execution. We want our algorithms: i) to discover patterns in large databases and have a completeness property that guarantees that all patterns of certain types have been discovered; ii) to have high performance and near-linear scaling properties; iii) to be particularly tailored to "mine" genetic data, so that to extract possible unknown genotype-phenotype correlations, to study modifier genes, and to select subgroups of patients for clinical trials, etc. With particular reference to the last point, machine learning techniques, both in the supervised and unsupervised framework, will be employed in order to extract significant relationships, eventually allowing auto-organization among data. Moreover, although successful in many applications, data mining poses special concerns for private data. Therefore, an integrated architecture must be devised that takes a systemic view of the problem, implementing established protocols for data collection, inference control, and information sharing (based on the European Union Privacy Directive mandate on privacy protection for data management and analysis systems).