Proper data analysis is crucial to obtaining valid and relevant results from an experimental system but is especially critical when assessing variations in mRNA expression of genes, the so-called transcriptome. Unlike DNA that is present in each cell throughout the life of the organism, RNA is transiently expressed and levels vary according to cell type, developmental stage, physiology, and pathology. Therefore, the quantification of RNA is context-dependent and inherently variable.
Extra vigilance must be exercised to ensure the data are technically accurate before they can be examined for biological relevance. It should always be remembered that mRNA expression does not necessarily correlate with protein expression and protein expression may not correlate with function. Regardless of the instrument used to perform quantitative real-time PCR, certain basic conditions should be met. The assay should be robust and fulfill all criteria for a good assay.
A standard curve using a defined template should result in a slope, coefficient of determination (R2), and y-intercept that demonstrate good efficiency, accuracy, and sensitivity. The baseline and threshold should be properly set. The amplification curves should demonstrate exponential amplification and be within the detection limits of the instrument. Proper controls should be included to monitor the contamination and sensitivity of the assay. Genomic contamination should be assessed, if applicable. Data should be reported in a manner that allows the variability of results to be seen. Proper statistical analyses should be applied to the data.
The interval when sample fluorescence reaches a predetermined threshold above background fluorescence is known as the Ct. This is often referred to as the crossing point or the Cp. A positive Ct value is caused by genuine amplification, but certain Ct values are not caused by genuine amplification, and some genuine amplification does not result in a Ct value. The Ct is determined by both the baseline and threshold settings that are selected.
The user can choose these configurations or approve the software’s default values. Values may be either set or adaptive. To compensate for slight well-to-well difference, the reporter signal is usually normalized to a reference dye by separating the raw fluorescence of the reporter by the fluorescence of the passive reference. This value is known as the Rn, or normalized reporter signal. When the background value is subtracted from Rn, the resulting value is known as delta (Δ) Rn, or the normalized fluorescence value adjusted for the background.
The copy number of the target of interest must be understood by using an absolute normal curve. This requires that the prototype be correctly quantified prior to curve generation. A sample of the target of interest contains exactly 2 x 1011 copies. The sample is diluted 10-fold eight times, down to 2 x 103 copies, and real-time PCR is done on each dilution with at least three replicates. The resulting standard curve associates each copy number with a specific Ct. The unknown samples’ copy number values are then determined by comparing them to this regular curve. The precision of the quantification is directly proportional to the consistency of the standard curve. In absolute quantification, consider the following:
- The template for standard curve generation as well as the method used to quantify that template is the foundation for the experiment. Pipetting accuracy for the dilution series is essential. Also, remember that real-time PCR sensitivity amplifies minute human error.
- Similar reverse transcription and PCR efficiencies for the target template and dilution series of the actual samples are critical.
How is the standard curve generated? Begin with determining the OD260 of the template sample. The amount of the template should be expressed in molecules. Convert the mass to molecules using the formula:
For example, if a synthetic 75-mer oligonucleotide (single-stranded DNA) is used as the template, 0.8 pg would be equal to 2 × 107 molecules:
What if there are no curves or strange curves? This condition must be addressed before any further analysis can take place. The first thing to verify is that the right dye layer or reporter has been assigned. This seems to be a straightforward proposition, but since no amplification curves or odd curves exist, this is the most possible interpretation. This issue is most common on multi-user instruments and where templates are used often.
For example, if FAM and SYBR reporters are used with the same well recognition, the spectra of FAM and SYBR Green may overlap, the amplification curves may appear respectable, and the Ct values may be indistinguishable. The variance can only be observed if the right dye layer is used. Another way to detect/verify this issue is to pick a well and examine it in the Raw Spectra view with the best match alternative enabled. The raw spectra curve for the well and the suit curve should be superimposed over each other.
If they aren’t, it’s a sign that the wrong reporter is being monitored (or there is contamination with another dye on the plate or in the sample block). What should be done to address this issue if it arises? To obtain the correct Ct values, choose the appropriate dye base, reassign the wells, and reanalyze. One method for reducing the possibility of using the incorrect dye layer is to reboot the data collection machine between individual runs, which resets template/software parameters to their previously allocated values.
If any appropriate amplification curves are found, the next parameter to investigate is the baseline condition. Each real-time PCR instrument has baseline-setting algorithms that use the fluorescence produced in a predetermined number of cycles at the start of the run to decide the background noise of the detector and reagents. Depending on the type of instrument used, there can be a constant number of cycles for all samples or flexible for each sample. Anomalies at the baseline level can be mild or serious. Examine the Amplification plot’s left-hand corner. Are there any ‘half-rainbows,’ or what seems to be a half-arc or hump between periods 1 and 10?
This arcs almost always begin above the threshold. In rare situations, full ‘rainbows’ affecting the whole curve can be observed. The ‘rainbows’ may appear under the threshold, and the Ct or curve may pass the threshold twice. Examining the Raw Spectra view or the Multicomponent view, which enables viewing of the raw fluorescence in a single well, will reveal whether or not there is actual amplification.
At what cycle does the lowest Ct appear?
If the lowest Ct is less than the upper limit of the baseline setting (or if the upper limit of the baseline setting is greater than the lowest Ct), the baseline should be updated. A common rule of thumb is to set the upper limit of the baseline 2–3 cycles less than the lowest Ct. Examining the amplification curves with the y-axis (Rn) set to a linear scale rather than a logarithmic scale is a reasonable way to decide where to set the baseline on certain machines. If the upper limit of the baseline is set very high, curves with very low Ct can appear below the baseline. Adjust the baseline until the longitudinal component of the curves matches the baseline. Correct baseline change could result in decent looking curves in the logarithmic view, with the rainbows gone.
Typically, the most noticeable result of changing the baseline would be on samples with the lowest Ct and the most template. However, limiting the number of cycles over which the baseline is measured can, in some rare situations, conflict with the identification of less abundant targets. In cases with extreme overloading of the template, it could be possible to exclude the offending samples from the study, as in the case of a regular curve, where values can be determined from different doses. This, however, has the potential to reduce production. If the questionable wells are essential to the experiment, such as the reference gene, it might be possible to replicate the experiment with less template, keeping in mind that a two-fold dilution of the template would only alter the Ct by a factor of 1. Because reference or standardization genes are so abundant, excessive gene expression is common.
The next critical move in data processing is to position the threshold correctly. When it comes to the baseline, each instrument uses a different type of a threshold-setting algorithm to determine this value. Typically, all real-time PCR analysis software measures the threshold value as 10 standard deviations from the baseline. As a result, before changing the threshold value, the baseline must be established.
An amplification curve must have four components: the history, the exponential phase, the linear phase, and the plateau. Set the threshold value within the exponential step of all amplification plots when viewed using the logarithmic scale for the y-axis to better change the threshold. If it is not possible to position the threshold in the correct location for all curves due to differences in the abundance of the various genes and/or the amount of fluorescence produced by a certain probe, then it is appropriate to examine sets of curves with different thresholds. Multiple thresholds are the exception rather than the rule, but they can be very useful when there is a low abundance mRNA that produces very low Rn values (Bustin and Nolan, 2004a).
How do you know where to position the threshold during the exponential phase? This is a point of contention, with others arguing that the lowest possible threshold is optimal. Some research program adjusts the threshold to provide the highest R2 for a regular curve. Others assume that choosing the point of highest curvature on a 2nd derivative curve (second derivative maximum (SDM)) yields the most precise Ct. In this case, varying the threshold setting has little impact on the calculated number of known template copies, showing that there is no single optimum threshold value.
However, a two-fold discrepancy in copy number could result in one Ct difference regardless of where the threshold is set within the exponential step of amplification, assuming efficiency is equal to 100%. Curve fitting, baseline, and threshold algorithms are not standardized across platforms, but the results may be identical. A study of PCR performance and y-intercept values on nine different platforms using the instrument’s default settings vs manually tuning baseline settings and thresholds. In most situations, the default configurations produce effects that are very similar to those obtained by manual adjustment.
Since a higher concentration of the standard was used and the default baseline was set too high, adjusting the baseline was particularly useful for certain instruments in this case. Since measured concentrations will differ slightly between platforms, when comparing quantitative results between platforms, a typical curve should be used. By the end of the day, the information is relative. The consistency of subsequent runs is perhaps the most significant consideration. Fine-tuning the baseline and threshold might not be as necessary if there are significant variations between test samples, but if two-fold or single Ct differences are required, careful adjustment becomes crucial.
The ΔΔCt approach contrasts experimental sample effects with a calibrator (e.g., untreated or wild type sample) and a normalizer (e.g., untreated or wild type sample) (e.g., housekeeping gene). Ct values for the gene of interest (GOI) in the test sample(s) and calibrator sample are now modified in relation to a normalizer (norm) gene Ct from the same two samples using this process. The fold difference in expression is calculated using the resultant ΔΔCt value.
The ΔΔCt approach requires that the normalizer and target gene have equal efficiencies. Of course, the obvious question is: what is a reasonable spectrum of deviation? To find out, use the same samples to build a regular curve for both the normalizer gene and the target gene of interest. For each dilution, the average ΔCt between the normalizer and target gene can be calculated. It is the continuity of the value through each dilution that counts, not the value itself.
The Ct discrepancy between the target gene in the test and calibrator samples is used in the conventional curve system of comparative quantification, which is normalized to the reference gene Ct values and corrected for minute differences in amplification quality. This approach requires a standard curve to calculate amplification efficiency for both the normalizer and the gene of interest, but it avoids the assumptions provided by previous techniques. The normalizer gene must be the same for all samples in the study for this approach to work.
High resolution melting (HRM) curve analysis
HRM analysis is a new closed-tube post-PCR approach for detecting SNPs, novel mutations, and methylation patterns that uses a novel, homogeneous melting curve. HRM research is more adaptive than conventional melt curve profiling, which monitors the temperature at which double-stranded DNA dissociates into single-stranded DNA. The Tm, or melting temperature, of the substance is this temperature. For extremely quick data acquisition and highly accurate thermal control and accuracy, a realtime PCR instrument with improved optical and thermal capability, as well as analysis tools, is needed for high resolution melting curve profiling.
Approximately 80–250 bp gene fragments are amplified using PCR in mixtures containing a high-performance double-stranded DNA (dsDNA)–binding dye in HRM analysis. The amplification materials are annealed and produce strong fluorescent after 40 cycles. The real-time PCR instrument gradually raises the temperature while continuously collecting fluorescence data from the amplicons as the HRM study starts. Since the fluorescent dye is released as the PCR products denature (or melt), their fluorescence can gradually decrease before the temperature reaches the Tm of the PCR product.
Multiplex real-time PCR analysis
Multiplexing is a method that involves analyzing multiple targets in the same real-time PCR reaction. A specific dye conjugated to the fluorogenic probe or primer pair specific for that target distinguishes each target. Multiplex reactions are often used to simultaneously amplify a normalizer gene and a gene of interest.
The number of targets that can be amplified in a given reaction is theoretically constrained only by the number of spectrally distinct dyes that are available, as well as the number of dyes that can be excited and detected by the real-time PCR instrument. Other experimental challenges remain, however: the various primer pairs and/or probes in a reaction must not overlap, and these primers and probes must share usable PCR components, such as dNTPs and thermostable DNA polymerase. Multiplexing has many distinct benefits, including being more time intensive to optimize:
- There would be less variation and more continuity. When a normalizer and a gene of interest are multiplexed in the same channel, well-to-well heterogeneity is eliminated, which may occur when the same two targets are amplified in separate (even adjacent) wells.
- Less reagent consumption equals lower costs. To evaluate the same number of targets, multiplexing necessitates fewer reactions.
In a multiplex reaction, the reporter fluorophores must be spectrally distinct such that the fluorescence signal from each is observed in a single channel. With compatible dyes, the excitation and emission optics of real-time PCR instruments will filter wavelengths such that less, if any, fluorescence is due to the incorrect fluorophore (such interference is also called cross-talk).
Similarly, as the number of probes being multiplexed rises, the quencher for each dual-labeled fluorescent probe becomes more significant. TAMRA dye, a popular fluorescent quencher for FAM dye, works by releasing the fluorophore’s energy at a different wavelength. Quenchers of this kind produce several signals at various wavelengths in a multiplex reaction, complicating filtering and potentially jeopardizing data integrity. Dark quenchers, such as NFQ or QSY quencher, emit energy as heat rather than fluorescence, lowering the total fluorescent history.