Summary of Outbrief
Dr. Simon Kasif presented the findings of the group that examined ways to use engineering-inspired algorithms and representations to understand biology, using language as a metaphor. He began by reviewing the progress made in cataloging the entire genetic code of entire living organisms, more than 100 to date. And just as the early developers of computer science struggled with explaining computation in the abstract with language-based ideas, scientists doing research on genomes face similar challenges in using language to describe complex events on various levels of scale.
The group heard from four presenters who discussed current developments in genome research:
Scientists can detect foreign DNA in microbial organisms. This capability could be useful in identifying—and countering—biological threats made against the United States and its allies. Very economical experiments are being developed to identify predictive DNA binding proteins in a mathematical way. Different levels of languages and scale are being used in describing biological systems. General languages are being used to describe protein-protein interactions and biological events.
Dr. Kasif concluded by outlining two courses of action. The first focused on what he described as "the computationally well-formed biological processes or questions." This involves building benchmarks via the speech [language] community to take the next leap in algorithms. A second challenge, one that he asserted was not as well-formed, involved the "virtual human" question and its implications on the pace of research and accompanying higher risks.
The group concluded that DARPA should continue its investment in following paths that answer the most risky biological questions such as developments in the immune system, human pathogens, virtual cells, or biological pathways that cut across organisms.
|
