Replicate analysis

When considering single-cell analysis and designing a real-time PCR assay, it does not follow that the more concentrated the DNA sample the more accurate the measurement. Many investigators may favor making a single measurement on the complete lysis product, rather than replicate measurements on multiple aliquots, either for data quality or cost reduction. It is essential when analyzing mtDNA, or any DNA, by real-time PCR to make replicate measurements on each sample/standard to have confidence in the data obtained. In the majority of cases the replicates will be similar, questioning the need and expense of replicate measurements. However, there are times when an outlier occurs and the inclusion of this in the analyzed data seriously affects the conclusions. When analyzing replicates these outliers can be identified and removed from the analysis. However, when making replicate measurements it is important to maintain the pipetted volume at a sensible value. If trying to replicate small volumes, a small percentage error in volume has a much greater effect on the final measured quantity. To this end we would recommend that the minimum final volume should be 5 ul, so the cell lysis volume should be adjusted to allow for this. If using a SYBR® Green reaction this may limit the investigator to two targets per cell before the starting quantity of DNA is too dilute to amplify. This raises the circumstances mentioned earlier when the benefits of a multiplex probe based real-time assay would enable the investigation of more targets per cell.

Table 11.7 Replicate vs. single measurements. A. Total (TOT) and wild-type (WT) mtDNA copy numbers measured in five single skeletal muscle fibers from a control individual. Four replicate measurements were made per fiber. Highlighted in bold are potential outlier values, approximately a factor of ten greater than the other three measurements; B. Average copy number values for both TOT and WT including all four values from each fiber, shown also is the calculated percentage deletion; C. Average copy number values for both TOT and WT excluding the potential outliers, shown also is the calculated percentage deletion. Exclusion of the potential outliers puts the average TOT and WT copy numbers more in agreement with each other, as expected in control fibers.

Copy number Average values

TOT WT TOT WT % Deln

Fiber 1

30600

162000

Fiber 1

32400

64600

49.9

32500

32200

Fiber 2

65400

30500

-114.4

31800

31400

Fiber 3

18700

19800

5.69

34600

32600

Fiber 4

28800

29000

0.6

Fiber 5

63300

28100

-125.2

Fiber 2

172000

28600

28800

27800

(C)

29300

33700

Average values excluding outlier

31700

31900

TOT

WT

% Deln

Fiber 1

32400

32000

-1.0

Fiber 3

14600

21100

Fiber 2

29900

30500

1.8

18400

19700

Fiber 3

18700

19800

5.7

20300

19300

Fiber 4

28800

29000

0.7

21300

19000

Fiber 5

26300

28300

7.3

Fiber 4

29200

30300

29200

28100

27700

28600

28900

28800

Fiber 5

24800 27300 26700 174000

25000 28600 31500 27400

In skeletal muscle fibers from control individuals we have investigated replicate versus single measurements. We performed two reactions per fiber using a single measurement for each reaction (i.e. measuring the percentage deletion). In control fibers, the copy number value obtained from each reaction should be identical and the calculated percentage deletion should be zero. The calculation of the ratio magnifies the observed error however, high percentage deletion values and large negative values can be seen (see % Deln values in Table 11.5). When the investigation was repeated making quadruplicate measurements for two reactions there was often a single replicate which stands out as an obvious outlier (Table 11.7). With all replicate values included in that analysis, the percentage deletion values can be seen to be more variable. Statistically, the outlier values can be legitimately removed if they fall outside of two standard deviations of the remaining three replicates. When this screening is applied we can see that the percentage deletion values lie much closer to the zero value expected.

Was this article helpful?

0 0

Post a comment