Automated Dose Response Curve Fitting Outpaces Manual Operation
AstraZeneca (Alderley Park, UK) has developed a customized data
processing application in Origin that can analyze up to 2,000 datasets
per hour, an improvement of 50 to 100-fold over the number that
scientists could analyze using previous methods.
The development of this application has given company scientists
the freedom to spend more time on other aspects of drug research,
as well as enabling AstraZeneca to perform high-throughput studies
in which thousands of test compounds can be evaluated in the same
Large Volume of Data
In the development of new prescription medicines, AstraZeneca researchers
perform concentration response studies to determine the biological
response elicited by varying doses of certain compounds. Much of
the data obtained from these studies exhibits a sigmoidal, or slanted
"S" shape when plotted (log concentration on the x axis,
measured response on the y axis). Several equations can be used
to fit such sigmoidal data. Typically the logistic equation is used.
In the past, the task of performing each curve fit for these studies
was tedious and time-consuming, mainly because there is so much
data involved. These drawbacks led AstraZeneca to consider developing
a customized application to automate each step of the curve fitting
It was determined that the customized application needed to have
the ability to handle large datasets, perform nonlinear curve fitting,
as well as create publication-quality plots.
After researching the available options and ruling out the possibility
of creating something from scratch, AstraZeneca (Alderley Park,
UK) decided to build upon Origin, a highly programmable graphing
and data analysis software package.
To implement the application in Origin, AstraZeneca utilized Origin's
built-in programming capability, as well as the ability to interface
custom DLLs with Origin's nonlinear curve fitter.
The application is designed to first import the desired data file
including plate numbers, compound identifiers, and response values
for each compound. The application then allows the researcher to
change settings, such as how the data will be analyzed, viewed,
For fitting using the logistic equation, the application includes
built-in parameter initialization code that vastly improves the
speed with which each fit will converge. The code determines the
approximate locations of the top of the curve, the bottom of the
curve and the IC50. To perform the analyses, the application is
designed so that all the researcher needs to do is press a button.
Once a dataset is analyzed it is plotted (Fig. 1) for review. The
researcher can then remove outliers, assign comments to individual
compounds or graphs, or sort compounds in a particular order, e.g.
by potency. When finished, the results can be saved as a single
file or uploaded to a database. In all, the application typically
handles curve fitting at a rate of 2,000 datasets per hour, enabling
very high throughput studies.
Doing More with Origin
Although AstraZeneca's biggest benefits have come from using Origin
to automate dose response data processing, the company has created
other custom Origin applications as well. Some of these automate
data handling for substrate dependence studies, ligand binding studies,
and enzyme inhibition experiments. These applications are typically
used to analyze fewer datasets but can compare curve fitting with
up to eight different models and allow more detailed analysis.
AstraZeneca, headquartered in London, is a major international
healthcare business engaged in research, development, manufacturing,
and marketing of prescription pharmaceuticals and the supply of
healthcare services. It is one of the top five pharmaceutical companies
in the world with healthcare sales in 2004 of $21.4 billion and
leading positions in cardiovascular, gastrointestinal, oncology,
central nervous system, and respiratory products. Products include:
Losec (omerprazole), the biggest-selling pharmaceutical of all time;
Nolvadex (tamoxifen citrate), the world's most prescribed breast
cancer therapy; and Seloken (metroprolol), the world's leading cardioselective
beta-blocker. AstraZeneca has more than 64,000 employees, manufacturing
sites in 20 countries, and 11 major research centers in 7 countries.
On the web: http://www.astrazeneca.com/