"Integrative Artificial Intelligence as a Key Ingredient of Systems Biology"
A full understanding of biological complex systems will only be achieved by human scientists working in collaboration with powerful AI software that can cognitively integrate the very broad range of biological data currently available. The Biomind AI Engine is a unique software system inspired by this vision. It aims to providing automated analysis of diverse biological data, and automated inference based on diverse biological information. Utilizing the Novamente "artificial general intelligence" framework, it integrates probabilistic reasoning with supervised learning (evolutionary programming, support vector machines), unsupervised learning (evolutionary programming, clustering) and automated text understanding. One distinctive aspect of the Biomind framework, in a data analysis context, is that "background knowledge" from biological databases (some of it drawn directly from the databases, some inferred from the databases using probabilistic inference) is used in the course of analyzing each particular experimental dataset. Currently the Biomind AI Engine is embodied in the commercial Biomind Analyzer product for gene expression data analysis, and is also being used by the author and his colleagues for broader-ranging research in bioinformatics and systems biology. Applications to date, conducted in partnership with several biological research teams, have involved inferring novel diagnostic rules for major diseases based on analysis of gene expression data and data regarding rare genetic mutations. The software has also been applied to the automated inference of genetic regulatory patterns from gene expression time series data; and to the inference of likely functions for as-yet unstudied genes. Current research focuses on automated inference of signal transduction pathways and other subtle patterns in genetic and proteomic systems.