|
Image informatics is a new area of data management
that allows researchers to mine scientific images of all types using advanced
image data storage, retrieval, mining and analysis capabilities.
Already proven in use by major pharmaceutical companies, this technology
is of particular value in proteomics research where the efficient management and
analysis of image data is critical in identifying patterns of protein expression
and then correlating these with specific outcomes.
For example, searching on a specific pattern of protein expression that
connotes a biomarker provides all incidences where this pattern has been seen,
such as in specimens from breast cancer patients.
Post-treatment samples can then be retrieved to correlate changes in this
pattern, such as down-regulation of a protein, that correspond to treatment
response.
Using Image Data to Accelerate
Proteomics Research
Proteomics research is amassing volumes of image
data. Researchers are studying
protein expression to track protein changes associated with various diseases and
as an indication of therapeutic response. The
objective is often to identify potential drug targets and to develop biomarkers
for certain disease states. The
most common research tool for protein isolation and studies of expression
remains two-dimensional (2D) electrophoresis gels, a procedure that separates
proteins along the two dimensions of charge and mass.
Protein characterization is typically performed by mass spectrometry.
A useful method of determining changes in protein expression resulting
from disease or in response to treatment, 2D gels have proven their utility
since their introduction 25 years ago. Advances in computer-aided analysis tools
have automated the process whereby researchers can quantify and compare one 2D
image to another in about one hour.
Lacking until recently, however, has been
technology to automate the process of searching for protein expression patterns
and gel analysis methods that are of sufficient sensitivity and accuracy to
enable searches of this type. Image
informatics technology accurately analyzes and extracts information from protein
spots on a 2D gel image, quantifying and storing visual content based on various
parameters including location, shape, size and intensity.
Researchers can now conduct searches for a specific pattern of protein
expression across hundreds or thousands of gels contained in a single, Oracle®
database.
The ability to store, retrieve, and mine protein
expression patterns found in 2D gels allows researchers to verify or disprove
hypotheses and to quickly determine the desirability of a compound, thereby
speeding the validation process. Image
informatics applications may be used to compare gels in order to identify
differentially expressed proteins, to gain insight into mechanism of action, or
to rapidly identify biological mechanisms or pathways without characterizing
individual proteins.
Protein expression data stored in one database can
be readily correlated with other experimental data including related image or
non-image data such as histology results, chemical entity structures, LD50
and ED50 information. Researchers
can create Image-driven Structure/Activity Relationships (ISAR) tables that
extract and present image data corresponding to biological, chemical and protein
activity. For example,
by starting with a set of 2D gel images that exhibit a pattern that confers
toxicity, related image and non-image data can be pulled together and viewed in
one screen. At a glance,
researchers can view a more complete set of data than possible previously and
gain valuable insights into cause/effect relationships.
In addition to boosting productivity by means of
new insights and more informed decision-making, image informatics provides an
open infrastructure for researchers to share data and collaborate across
experiments and laboratories. Being
able to access images along with the expert’s opinion facilitates the advance
of appropriate compounds, no matter where in the organization they have been
tested or created. As
pharmaceutical companies partner with biotechnology companies in greater
numbers, new database management systems must prove sufficiently robust to
accommodate multidisciplinary collaborations with selective viewing privileges
for proprietary data.
Use of image informatics can advance proteomics
research at every stage – from hypothesis, experiment design and gel
characterization to identifying protein expression patterns, understanding and
validating mechanism of action and therapeutic response.
This relatively new technology helps researchers exploit image data
throughout the R&D process in ways that were not previously possible.
Researchers can enlist image informatics to design and conduct
retrospective and prospective evaluations, to identify mechanism of action and
pathways, and to make better use of enterprise legacy data.
Image data can now be integrated much more fully into the decision-making
process on a routine basis, serving as a valuable source of insight and
significantly accelerating the drug discovery and development process.
The impact of proteomics research is expanding
rapidly and affects nearly every stage of the pharmaceutical and biotechnology
R&D process. As proteomics
research advances, research techniques and methods used to make decisions about
the data it provides must also advance, becoming faster, more accurate and more
integrated. The explosion in image
data in particular has created a pressing need for improved technologies for
capturing, storing, mining and analyzing image data. Analysis tools designed to compare one gel to another no
longer suffice; proteomic researchers need technology capable of performing
rapid searches for protein expression patterns, of enabling sharing of data
between labs and company departments, and of integrating images with other,
related data. Image informatics
offers researchers, research teams and project teams a mechanism to use image
data throughout the entire discovery and development cycle in a way that is far
more efficient than was previously possible – opening up new possibilities for
insight.
The articles and opinions expressed in the Editorial
Corner are solely those
of the author and do not necessarily reflect the opinions of the
Proteome Society management and/or membership.
|