One thing that all of these applications of qPCR have in common is that they generate a great deal of data, and that computerized statistical methods are necessary to sort it out. When data analysis is carried out by a software package, it is often opaque to the scientist, and so it is not always apparent when things need to be improved.
When Terry Hyslop, Ph.D., director of biostatistics for Thomas Jefferson University (www.jefferson.edu), embarked on the data analysis for a clinical trial of guanylyl cyclase c as a prognostic biomarker for colon cancer, she noticed something interesting about the data coming out of their qPCR assays.
“We got the machine here and started seeing data coming out of it. The way the q process happens, it does not take into account the experimental realities. As a statistician, I know there’s a lot of data and assumptions in the process not being utilized. We found that if we incorporated these characteristics, the model provided a much more precise answer in terms of reducing the variability of what’s measured,” she explains.
Dr. Hyslop and her colleagues developed an algorithm, creating a mathematical curve to fit all of the data coming out of the machines. Using this algorithm, they reduced the noise in the experiment and lowered the error rate by 30–60%. Additionally, by using a newer quantification process, they eliminated the need to run a set of standard curve samples in each plate because the standards are internal to the experiment. The algorithm can add up to significant savings in the design and execution of new studies.
“When you go to plan a new study, the variability won’t be as high, so you won’t need as many patients in the study to answer the same question,” says Dr. Hyslop.
Unlike his colleagues at the conference, George Weinstock, Ph.D., associate director, Genome Center, Washington University, is not presenting an improvement or new application of qPCR. Rather, he is presenting a method for targeted genome sequencing that avoids the use of all forms of PCR. The method is an alternative to developing PCR primers, amplifying a particular region, and then sequencing the amplified fragments. It is a chip-based oligonucleotide array, utilizing 454 Life Sciences; a Roche Diagnostics
This is an important reminder that one should not be married to any single method or technology, and that a good scientist uses the best tools available even if that means moving out of one’s comfort zone. Researchers will continue to find ingenious new applications for qPCR, especially in combination with exciting next-generation tools in related disciplines.