Big Mechanism, Solicitation Number: DARPA-BAA-14-14
Reponse Data: Mar 18, 2014 12:00 pm Eastern
From the solicitation:
Here is one way to factor the technologies in the Big Mechanism program:
- Read abstracts and papers to extract fragments of causal mechanisms;
- Assemble fragments into more complete Big Mechanisms;
- Explain and reason with Big Mechanisms.
Here is a sample from Reading:
As with all natural language processing, Reading is bedeviled by ambiguity [5]. The mapping of named entities to biological entities is many-to-many. Context matters, but is often missing; for example, the organism in which a pathway is studied might be mentioned once at the beginning of a document and ignored thereafter. Although the target semantics involves processes, these can be described at different levels of detail and precision. For example, “β-catenin is a critical component of Wnt-mediated transcriptional activation” tells us only that β-catenin is involved in a process; whereas, “ARF6 activation promotes the intracellular accumulation of β-catenin” tells us that ARF6 promotes a process; and “L-cells treated with the GSK3 β inhibitor LiCl (50 mM) . . . showed a marked increase in β-catenin fluorescence within 30 – 60 min” describes the kinetics of a process. Processes also can be described as modular abstractions, as in “. . . the endocytosis of growth factor receptors and robust activation of extracellular signal-regulated kinase”. It might be possible to extract causal skeletons of complicated processes (i.e., the entities and how they causally influence each other) by reading abstracts, but it seems likely that extracting the kinetics of processes will require reading full papers. It is unclear whether this program will be able to provide useful explanations of processes if it doesn’t extract the kinetics of these processes.
An interesting solicitation. Yes?
I thought it was odd though that the solicitation starts out with:
DARPA is soliciting innovative research proposals in the area of reading research papers and abstracts to construct and reason over explanatory, causal models of complicated systems. Proposed research should investigate innovative approaches that enable revolutionary advances in science, devices, or systems. Specifically excluded is research that primarily results in evolutionary improvements to the existing state of practice. (emphasis added)
But then gives the goals for 18 months in part as:
- Development of a formal representation language for biological processes;
- Extraction of fragments of known signaling networks from a relatively small and carefully selected corpus of texts and encoding of these fragments in a formal language;
Aren’t people developing formal languages to track signaling networks right now? I am puzzled how that squares with the criteria:
Specifically excluded is research that primarily results in evolutionary improvements to the existing state of practice. ?
It does look like an exciting project, assuming it isn’t limited to current approaches.
[…] this talk about medical research reminds me of the Big Mechanism DARPA. Assume the research data on pathways is no better or no worse than mapping genes to diseases, […]
Pingback by Medical research—still a scandal « Another Word For It — February 9, 2014 @ 5:45 pm