Getting to Green - Research Progress on Environmental Health and Safety Aspects of Engineered Nanomaterials - page 59

scientific discussion of a particular model, its submodels, and algorithms; comparisons with similar
models; and a means of establishing provenance concerning both model development and authorship.
During model development (which accesses public data) the collaborative spaces should be open,
permitting easy access for annotating and curating model development, the model's theoretical
underpinnings, numerical methods and algorithms, and model validation. Access may be restricted for a
number of reasons, but provision should be made for eventual publication of the model because scientific
publication would be a primary motivator for placing it in the knowledge commons and would allow
faster model improvement and adoption through an open-development infrastructure. As noted above,
open development would be particularly suited to risk models in that modeling risk involves uncertainty
propagation, whether the uncertainties arise from models, data, or expert opinion. Through use of the
models to focus resources on reducing the largest uncertainties, the reproducibility of risk estimates could
be improved systematically.
The use of virtual collaborative environments would also be key for methods development,
creating a single focus for a method—its documentation and range of validity, accompanying video for
adding detail or providing training, current instrumentation and later improvements, links to data obtained
from the method and links to data and models derived from the data, annotation on the method and
datasets, information on sample preparation and controls for different ENMs, and metadata and
information regarding method curation and provenance. The primary advantages of the collaborative
environments for analytic methods would be a common focus for all aspects of method development,
robustness testing and capture of sensitivity data, interlaboratory testing and data capture, use of reference
materials for calibration, suggestions for improvements and extensions, method revision and retesting,
and provenance concerning all uses of the data. Standard methods could be developed, validated, adapted,
improved, and revised on an abbreviated timescale while linkage to all raw, derived, and modeled data
related to that method, its instrumentation, and sample preparation procedures is provided.
Virtual collaborative environments would also accelerate the development of nanomaterials. The
collaborative environment could focus on a particular ENM designated by a production lot number;
document the production, separation, and purification processes used and any initial characterization of
the lot's properties; and create a data aggregation point for all uses of the particular nanomaterial, how
samples were prepared, what methods were used, and whether the method's data were associated with
any models or modeling efforts. Sample history could also be recorded; this would provide data necessary
for both informal and formal interlaboratory testing of the materials with different methods. As data from
different researchers using different methods are accumulated, comparisons can be made with greater
validity because there would be a basis for “apple to apple” 2 comparisons, given the association among all
samples that have common parentage. In addition, informed decisions concerning the structure or
distribution of structures of particular ENM samples would be possible, and structural models of the
samples could be deposited in a repository, such as the Collaboratory for Structural Nanobiology, for use
in developing detailed predictive models of ENM effects in different environments. Collaboration spaces
would also support aggregation of data on ENMs from different lots or from similar materials. Analysis
of those data would allow correlation of ENM sample structures with their properties and effects and aid
in formulating hypotheses of possible underlying mechanisms.
Perhaps the most important effect of the knowledge commons is the creation of a new literature
based primarily on data from the application of validated methods to identified lots of nanomaterials. Raw
data would be linked to derived data—whether on nanomaterial structure, their properties, or their effects
in different experimental tests and environments—and to data from appropriate predictive and structural
models. The correlation of data on ENM lot, structure, properties, and effects would help in the creation
and incremental improvement of an evidence-based nomenclature and ontology that are consistent with
known structural, experimental, and modeling data and that can be used to organize and track the use,
2 The use of the phrase “apple to apple” comparisons conveys the importance of comparing sufficiently similar
nanomaterials in studies (including such information as the material size, physical and chemical structure and
properties, purity, and processes used to manufacture, store, and prepare the materials for analyses).