Does industrialized intellect embodied in robot scientists sound far fetched? They’re already here.
A team lead by Professor Ross King has brought together multiple technologies to create prototype robot scientists. One key factor has been the development of artificial intelligence technology into the globally distributed collaborative technology known as the Semantic Web.
The scientist robots have been built to automate all aspects of the scientific discovery process. They generate hypotheses, design and carry out the experiments, analyze and interpret the resulting data, and the robots can use the results of their earlier experiments to generate new hypotheses and to repeat the whole cycle.
Professor King’s team provides several compelling reasons why robot scientists will be an essential addition to the laboratory including improving the accuracy and reliability of experiments, fully formalizing science reporting, increasing the pace of discovery, and removing tedious and repetitive tasks from the human scientist. And they created two scientist robots to prove it.
Adam was their first scientist robot. He was designed to carry out experiments to study functional genomics in yeast. There are enzymes that are known to exist in yeast but their corresponding genes remain unidentified. Adam identified the genes encoding some of these enzymes (see The Automation of Science” published April 3, 2009 in Science).
Among the technologies that Adam used were a comprehensive logical model of yeast metabolism, the Kyoto Encyclopedia of Genes and Genomes (KEGG), and a custom built ontology for the logical description of experiments. The team has recently been collaborating on “the most complete reconstruction of yeast metabolism to date,” which should help boost Adam’s capabilities. The model is being built using Systems Biology Markup Language (SBML) and is publicly available at YEASTNET (see “Further developments towards a genome-scale metabolic model of yeast” published October 28, 2010 in BMC Systems Biology).
Eve is the team’s second robot scientist. Eve is being developed to carry out drug-screening and design. Her tasks will include running mass screenings, hit verification, and hypothesis-driven targeted screening.
Eve will use Drug Discovery Investigation (DDI), which is a new ontology that is being developed collaboratively with the drug discovery and development community. DDI will define the principle entities and relations in the drug discovery pipeline’s research and development phase. DDI is being implemented using Web Ontology Language (OWL) and leverages many existing ontologies (see “An ontology for description of drug discovery investigations” published March 25, 2010 in Journal of Integrative Bioinformatics).
Robot scientists and the industrialization of intellect is just one example of why I believe the Semantic Web is the defining new technology for the twenty-first century.
This blog post is based on the following papers (from most recent):
“Further developments towards a genome-scale metabolic model of yeast” published October 28, 2010 in BMC Systems Biology.
“An ontology for description of drug discovery investigations” published March 25, 2010 in Journal of Integrative Bioinformatics.
“Towards Robot Scientists for autonomous scientific discovery” published January 4, 2010 in Automated Experimentation.