ABSTRACT
The most notable, recent developments in quantitative polymerase chain reaction (qPCR) technology have been achieved by improving technological elements of PCR. Modern cyclers have a more uniform temperature over the plate, new DNA polymerases are able to synthesize across more complex secondary structures of DNA, and novel detection chemistries are capable of more specic amplication and monitoring of the amplied DNA. Jointly, these advances have provided users with solid instrument performance, readily achieving precision as low as 0.25 Cq among PCR technical replicates.1 In general, the issue of precision that is achievable for technical replicates has been emphasized largely by the instrument manufacturers, often ignoring the importance of biological heterogeneity among subjects of research and the effect of the sample preprocessing steps (see Chapter 1). This may have contributed to a perception that quantication of gene expression by real-time PCR (qPCR) is challenged by technical imperfection at the PCR level. A recent study in
17.1 Introduction ........................................................................................................................245 17.2 NeuroSearch Example Data ...............................................................................................246 17.3 Exploratory and Conrmatory Approach in Statistics ...................................................... 247 17.4 Statistical Tests and Condence Intervals .........................................................................248 17.5 Power Analysis ...................................................................................................................248 17.6 Traditional Visualization Techniques ................................................................................249 17.7 Alternative Visualization Technique and the Condence Interval of the Difference
between Means .................................................................................................................. 251 17.8 Interpretation of the Visualization ..................................................................................... 252 17.9 Multiple Testing ................................................................................................................. 253 17.10 Minimizing Spurious Effects and Confounding Factors ...................................................254 17.11 More Complex Experimental Design ................................................................................ 255
17.11.1 Pilot Study ............................................................................................................256 17.12 Conclusions ........................................................................................................................256 References ...................................................................................................................................... 257
which the Cq precision of replicates is tested2,3 revealed that the PCR component of a gene-expression experiment yields, by far, the highest precision when compared to all the preceding steps of sample acquisition and processing. Hence, there is a greater potential to reduce data variability and improve accuracy by putting more focus on improving the experimental design. The parameters within the experimental design that may benet from additional attention include the required number and distribution of biological subjects within experimental groups and the number of replicates employed for individual sample-processing operations and the points of the workow at which these are taken. In our experience, strictly adhering to the use of appropriate statistics that was dened prior to taking the pipette in hand further increases the overall validity of results. The power of a statistical test depends on several factors, including (1) the statistical signicance criteria, (2) the sample size, (3) the size of experimental effects, and (4) the magnitude of variability in experimental measurements. The aim of this contribution is to provide the reader with concise guidelines for design and analysis of gene-expression experiments and to advice on how to undertake appropriate analyses. We will present ways to visualize data to highlight all the factors that demonstrate the power of the statistical test. Throughout all stages, the objective is to achieve the strongest possible biological inference.
