Klein, Wallin, and Buoncristiani, "Addressing Ambient Temperature Variation Effects on Sizing Precision of AmpFlSTR® Profiler Plus™ Alleles Detected on the ABI Prism® 310 Genetic Analyzer"
January 2003 - Volume 5 - Number 1
Research and Technology
Addressing Ambient Temperature Variation Effects on Sizing Precision of AmpFlSTR® Profiler Plus™ Alleles Detected on the ABI Prism® 310 Genetic Analyzer
Sonja B. Klein
Jeanette M. Wallin
Martin R. Buoncristiani
California Department of Justice DNA Laboratory
Abstract | Introduction | Materials and Methods | ABI Prism® 310 Genetic
Analyzer Sample Preparation and Electrophoresis | Data Analysis | Results: GS-500 and 400HD | Size Deviation with Ambient Temperature | Global
Southern | GS-500 and ILS-600 | Discussion | References
Early studies have established the Local Southern algorithm as a precise tool for sizing DNA fragments. As a result, the Local Southern algorithm of the PE Applied Biosystems' software, GeneScan® Analysis (PE Applied Biosystems, Foster City, California), is the manufacturer's recommended method for sizing short tandem repeats (STRs). However, this recommendation is made with the warning that size estimates may be imprecise if any of the standard fragments run anomalously. Specifically, the GeneScan®-500 (GS-500) internal standard fragments of 250 and 340 bases in length run anomalously under nonoptimal conditions on the ABI Prism® 310 Genetic Analyzer (PE Applied Biosystems, Foster City, California).
The California Department of Justice DNA Laboratory currently uses the GS-500-size standard without the 250-base standard assigned and the Local Southern method to size AmpFlSTR® Profiler Plus™ alleles. However, even with the manufacturer's recommended instrument running conditions, studies in this laboratory demonstrate that ambient temperature variation over the course of a 310 run can result in anomalous migration of GS-500 standard fragments. When ambient temperature varies, a simple analysis method change can improve precision.
This study suggests that the Global Southern method may provide improved precision over the Local Southern method when using the GS-500 internal standard with the ABI Prism® 310 Genetic Analyzer. In addition, this study shows that precision for fragments greater than 300 bases is further improved by excluding the 340-base GS-500 fragment in conjunction with using the Global Southern method. When ambient temperature shifts occur, this sizing method change should reduce the number of sample reruns necessary.
The ABI Prism® 310 Genetic Analyzer is one of the more commonly used DNA analysis instruments in forensic laboratories. It is an automated, single-capillary electrophoresis instrument with laser-induced fluorescence detection. The California Department of Justice DNA Laboratory uses the ABI Prism® 310 Genetic Analyzer for the separation and detection of STRs amplified with the AmpFlSTR® Profiler Plus™ PCR amplification kit (PE Applied Biosystems, Foster City, California). The fluorescently tagged STR products are coinjected with the GS-500 ROX-internal standard and sized using an interpolation algorithm for Local Southern in the GeneScan® Analysis software. The allelic ladder containing the common STR alleles is also injected with the GS-500 ROX-size standard. The Genotyper® software (PE Applied Biosystems, Foster City, California) performs automated allele calling, comparing the sizes generated from one injection of the allelic ladder to those generated from sample injections. Sample fragment sizes within ±0.5 bases of the corresponding allelic ladder fragment are assigned the appropriate allele designation. The one-base window for genotyping is based on the assumption that single-base sizing precision may be routinely achieved (Lazaruk et al. 1998). In order to obtain single-base sizing precision, standard deviations should be within approximately 0.16 bases so that given three deviations, 99.7 percent of the sizes will statistically fall within 0.48 bases from the mean (assuming that the sizes are generated by random error). If standard deviations greater than approximately 0.16 bases are observed, genotyping accuracy may be compromised. Although these STRs are tetranucleotide repeats, one-base microvariant alleles do occur at some loci; therefore, single-base sizing precision is desirable.
One critical factor affecting precision is the electrophoresis running temperature. Changes in running temperature affect the viscosity of the polymer matrix and the sieving of DNA fragments. It can also affect DNA secondary conformation. Because the entire length of the capillary on the 310 Genetic Analyzer is not insulated, ambient temperature fluctuations that occur can affect the run temperature, at least in the exposed portions of the capillary. The manufacturer recommends that the ambient temperature not vary by more than ±2ºC during a run (PE Applied Biosystems 2000). As shown previously (Rosenblum et al. 1997), this specification is important for fragments of the GeneScan®-350 (GS-350)- and GS-500-size standards. The 250-base fragment is particularly affected by run temperature changes, which cause it to migrate anomalously. Hence, the 250-base fragment is not included in the generation of size standard curves (PE Applied Biosystems 2000). Ambient temperature fluctuation may also cause the 340-base fragment to migrate anomalously, although to a lesser degree. However, most forensic laboratories include the 340-base fragment when generating size standard curves.
The degree to which anomalously migrating size standard fragments influence the curve should be dependent upon the size method employed. Elder and Southern (1983) found local algorithms to be more precise for sizing DNA polymers than global algorithms, but they also found precision was greatly reduced when there were sequence differences between standards and unknowns. The effects of such sequence differences should minimize when using global methods. Most forensic laboratories, including the California Department of Justice DNA Laboratory, use the Local Southern method, as recommended by the manufacturer, although size estimates may be imprecise if any of the standard fragments run anomalously (PE Applied Biosystems 1998). The size range for the more common alleles of the AmpFlSTR® Profiler Plus™ loci is approximately 100-345 bases. While the specifications of the 310 Genetic Analyzer polymer, Performance Optimized Polymer-4™ (POP4™) (PE Applied Biosystems, Foster City, California), are to provide single-base sizing precision up to 250 bases, laboratories have reported such precision up to approximately 350 bases on this platform when using Local Southern sizing (LaFountain et al. 2001; Lazaruk et al. 1998; Moretti et al. 2001). Furthermore, routine analyses in this laboratory using AmpFlSTR® Profiler Plus™ with GS-500 and Local Southern indicated single-base sizing precision. However, periodic runs have shown losses in precision, presumably due to fluctuations in ambient temperature.
|Figure 1 Size Standards Click to enlarge image.|
The following experiments were performed to examine the cause of the suboptimal precision with conventional methods and to investigate alternative approaches that may offer single-base sizing precision more routinely. Alternative approaches tested included analyses omitting the 340-base fragment, an investigation into the use of Global Southern, and an evaluation of the size standards GeneScan®-400HD (400HD) (PE Applied Biosystems, Foster City, California) and Internal Lane Standard 600 (ILS-600) (Promega Corporation, Madison, Wisconsin). Each of the size standards are illustrated in Figure 1.
Twenty-four microliters of high-deionized formamide (Hi-Di™ formamide, PE Applied Biosystems, Foster City, California) were combined with 1µl of GS-500, 400HD, or ILS-600. Next, 1.5µl of Profiler Plus™ allelic ladder (PE Applied Biosystems, Foster City, California) was added to each tube containing Hi-Di™ formamide and GS-500, Hi-Di™ formamide and 400HD, or Hi-Di™ formamide and ILS-600. Samples were heat-denatured at 95°C for three minutes, snap-cooled on ice for three minutes, then placed in a 48-well autosampler tray for automatic injection in a 47 cm (36 cm length to detector) 50 µm inner diameter capillary on the ABI Prism® 310 Genetic Analyzer. Samples were electrokinetically injected for five seconds at 15 kV, then run at 15 kV for 26 or 28 minutes in POP4™ at 60°C (Run Module GS STR POP4™, F) with GeneAmp 10X Buffer (PE Applied Biosystems, Foster City, California) at 1X. Ambient temperature during the runs was monitored by placing a temperature gauge (Omega Engineering, Stamford, Connecticut) adjacent to the 310 Genetic Analyzer. Repeat injections from each tube of ladder were performed, alternating between GS-500 and either 400HD or ILS-600.
All sample files were analyzed with the ABI Prism GeneScan® v2.1 or v3.1 software using both the Local and Global Southern sizing methods. The GS-500 250-base fragment was excluded in the size standard assignment of all analyses. In addition, some data analyses also excluded the 340-base fragment of the GS-500.
|Figure 2 Standard deviations for each allele of 50 Profiler PlusT allelic ladder injections sized with GS-500 (¨ ) and 50 sized with 400HD (¨). Click to enlarge image.|
Profiler Plus™ allelic ladder was injected 50 times with GS-500 and 50 times with 400HD. The injections alternated between GS-500 and 400HD so that conditions for each internal standard would be as similar as possible. The ambient temperature over the two-day run ranged from 16°C to 20°C. Precision was estimated by calculating the standard deviation at each of the 118 alleles after sizing with the Local Southern method. The standard deviation for each allele size is shown in Figure 2. Most notable is the increased standard deviation of allele sizes greater than 300 bases sized by GS-500. The precision between GS-500 and 400HD appears similar otherwise, with GS-500 generating slightly better precision for fragments between 175 and 290 bases in length. However, the loss of precision with the GS-500 relative to the 400HD for Profiler Plus™ alleles greater than 300 bases suggests that precision is limited by the GS-500-size standard. Based on the increased standard deviations of the Profiler Plus™ alleles with the GS-500, the 340-base fragment was suspect. When the allelic ladders were resized excluding the 340-base standard fragment, the precision improved and the standard deviations for fragments above 300 bases were similar to those generated with the 400HD.
|Figure 3 The size deviation of each allele from the average is plotted with increasing size (observed minus average). Click to enlarge image.|
To examine the effects of ambient temperature on GS-500 precision, size deviations from the allele averages were compared to the measured ambient temperature at the start of each injection (Figure 3). The ladders injected at the highest and lowest ambient temperatures sized furthest from the average, demonstrating a relationship between calculated size and temperature. In other words, the larger the ambient temperature shift, the larger the size deviation—specifically, alleles sized larger than the average when the ambient temperature decreased and smaller when the ambient temperature increased. Therefore, at lower ambient temperatures, the alleles migrated slower than the GS-500 fragments and at higher ambient temperatures, the alleles migrated faster than the GS-500 fragments. In addition, the ambient temperature changes affected the 340-base fragment more than the other assigned standard fragments. Thus, there was the increased deviation for alleles larger than 300 bases.
|Figure 4 The data point (retention time) difference between lowest (16.5°C) and highest (20°C) ambient temperature injections is plotted versus size for each fragment (FAM, JOE, and NED are allelic ladder dye labels; ROX is the GS-500 dye label). Click to enlarge image.|
To illustrate the migration differences among and between the Profiler Plus™ alleles and the GS-500 fragments with ambient temperature change, fragments from the highest and lowest temperature injections were compared by data point position (retention time). The data point position for each fragment of a single 20°C injection was subtracted from the data point position of the corresponding fragment in a single 16.5°C injection (Figure 4). Data point differences increased linearly with fragment length as a result of the longer retention times for the larger fragments. High correlation coefficients reflect migration consistency in fragment spacing, even though fragment retention times changed with temperature. While the correlation coefficient was high for each dye-labeled set of fragments, the GS-500 fragments exhibited the lowest linear regression determination coefficient (R2 = 0.9951), with the 250- and 340-base fragments visibly offset, presumably because of sequence content variability. These two fragments shifted less with ambient temperature change than the other fragments, resulting in the smaller data point difference. Thus, the linear regression determination coefficient changed from 0.9951 to 0.998 when the 250- and 340-base fragments were excluded. Additional GS-500 fragments appeared to slightly deviate from the linear trend (160- and 300-base fragments). This observation implies additional GS-500 fragments, to a lesser extent, may migrate anomalously with increased ambient temperature variations and that the GS-500 fragments are not proportionately affected by temperature change as the ladder alleles.
Figure 5 Standard deviations for each allele of 50 Profiler PlusT allelic ladder injections with GS-500 (Δ) and 50 with 400HD (¨) sized using Global Southern. Click to enlarge image.
Because at least some of the GS-500 fragments ran anomalously, the Global Southern sizing method was applied to the same data to test if it improved precision. Data reanalysis using Global Southern, both with and without the 340-base fragment, improved precision relative to analysis with the Local Southern method. The highest precision was obtained using the Global Southern method with the exclusion of the 340-base fragment (Figure 5). The largest standard deviation observed with this method was 0.19 base, and only six allele sizes had standard deviations greater than 0.16 bases. For fragment sizes greater than 300 bases, Global Southern greatly improved precision over Local Southern with the inclusion of the 340-base fragment, from close to 0.30 base standard deviation to less than 0.10 base standard deviation. The precision trend changed from a rise in standard deviations for fragments above 300 bases to an overall lower level of standard deviations across all fragment sizes. This consistency across fragment sizes demonstrates how the Global Southern method minimizes size deviations caused by anomalous migration of internal standard fragments. When using the Global Southern method, the GS-500 produced better precision than the 400HD. The precision obtained when using 400HD with Global Southern was relatively unchanged compared to that with Local Southern (Figures 2 and 5). This is probably because there are no anomalously migrating 400HD fragments (PE Applied Biosystems 2000).
|Figure 6 Standard deviations for each allele of 50 Profiler PlusT allelic ladders coinjected with GS-500 (¨) and 50 coinjected with ILS-600 (•). Click to enlarge image.|
To examine sizing precision using ILS-600, a total of 100 Profiler Plus™ allelic ladder injections were performed, this time alternating between GS-500 and ILS-600 sizing standards. The ambient temperature during the run ranged from 20°C to 23°C. The precision generated by both standards was well within a single base, with every allele sizing less than 0.13 base standard deviation. However, there were still notable differences in precision with GS-500 when sizing with Local Southern versus Global Southern. Sizing precision of the ILS-600 was consistent across the range of allele sizes with a standard deviation range of 0.03 to 0.06 base (Figure 6). The ILS-600 fragments apparently migrated similarly to the Profiler Plus™ alleles with ambient temperature change. Sizing precision of the GS-500, although consistent up to 290 bases with a standard deviation range of 0.04 to 0.07 base, increased in standard deviation to a maximum of 0.12 base for the largest alleles. Although each allele sized within the Genotyper® ± 0.5 base window and minimal ambient temperature variation was observed during the run, the 340-base standard still apparently migrated anomalously. When the GS-500-sized allelic ladder injections were reanalyzed with the Global Southern sizing algorithm excluding the 340-base standard, the precision again improved, resulting in a maximum standard deviation across all fragment sizes of 0.07 base.
STR sizing precision is dependent on ambient temperature control when using the 310 Genetic Analyzer with the internal standard GS-500. This study showed that allele sizes were reproducibly smaller when ambient temperature increased and larger when ambient temperature decreased. The basis of this phenomenon appears to be that as ambient temperature increases, Profiler Plus™ alleles exhibit an increased mobility relative to GS-500 fragments. The phenomenon is most pronounced with the 250- and 340-base fragments of the GS-500 sizing standard and is probably attributed to the varying sequence content between the size-standard fragments as well as the DNA sequences that are actually measured. In contrast to GS-500, the ILS-600 standard consists of same-sequence subsets. Therefore, when the ILS-600 is used on the 310 Genetic Analyzer, it generates better sizing precision than the GS-500 sizing standard. It appears that the number of internal fragments assigned is not a critical factor in precision. The 400HD has 18 fragments between and including 100 to 400 bases, whereas GS-500, excluding the 250 and 340, has only eight fragments between and including 100 to 400 bases. Lazaruk et al. (1998) also found no advantage to using the more dense 400HD over the GS-350. More critical than the number of internal fragments or the spacing between them is the similar migration changes of alleles and standards with changing running conditions. The sizing differences are minimized when sequence content is similar.
Size variation is due, in part, to temperature fluctuations in the capillary itself. Although the majority of the capillary on the 310 Genetic Analyzer is temperature-controlled by a heat plate, small regions of the capillary are exposed to ambient air where fragment migration is likely to be affected by ambient temperature changes, particularly at the cathode end. In addition to the total ambient temperature change, the duration at each temperature also seems to play a role in fragment migration. Although the first (GS-500/400HD) and second (GS-500/ILS-600) runs had similar absolute ambient temperature fluctuation ranges, the second run had much better precision. In run two, the duration at which the temperature was low or high was very short (less than one hour), whereas for run one, the duration was longer (up to four hours). This presumably provided more time for the DNA fragments to be affected by conformational change or the polymer sieving properties. The range for run two was also at an overall higher temperature than for run one; this may have also played a role in the precision.
Because one sizing of allelic ladder is generally used to genotype a large number of samples, it is best to consider the sizing precision over the entire duration of the run. Ideally, the standard deviation for every allele size in a given run should measure to 0.16 base or less; this would statistically mean 99.7 percent of the allele calls size within ± 0.48 bases (or three standard deviations) of the mean. This level of precision can be achieved in a number of ways. Ideally, the ambient room temperature could be strictly controlled. Alternatively, the GS-500 internal lane size standard may be replaced by other standards available that contain same-sequence subsets. Excellent sizing precision with the ILS-600 standard was observed by Pawlowski and Maciejewska (1999), as well as in this study. However, this study also found the correction for spectral emission overlap of the CXR-labeled fragments less effective than that generated with the ROX-labeled fragments (Filter Set F). Pull-up in the range of 2.0-3.5 percent in the NED panel occurred when the CXR matrix was applied (NED pullup peak relative fluorescence unit (RFU)/CXR peak RFU). Finally, and perhaps the least involved approach, is the optimization of the size calling method performed.
This study compared the Local and Global Southern methods for the sizing of Profiler Plus™ alleles. Briefly, both methods use the equation L = [c/(m-m0)] + L0 to describe the reciprocal relationship between the mobility, m, and the length, L0, of the standard fragments (Elder and Southern 1987). However, the Local Southern method averages two fragment values generated by two curves each of three standard fragments closest to the unknown, whereas the Global Southern generates one fragment value by calculating a best fit line of all standard fragments selected (GeneScan® Analysis Software User's Manual). Thus, Global Southern is the better sizing algorithm to use to minimize imprecision due to anomalously migrating standard fragments. However, this also results in fractionated values for the size-standard fragments. For example, when the Global Southern sizing algorithm is used, values such as 74.8 and 400.1 bases are shown for the 75- and 400-base internal lane fragments after analysis. Consequently, when using the Global Southern sizing method, the size call range in analysis parameters of GeneScan® should be increased from minimum 75, maximum 400 to approximately minimum 74, maximum 402 if the 75 and 400 fragments are to be included in every data set.
Anomalous migration as a result of ambient temperature fluctuation was observed with GS-500 fragments on the 310 Genetic Analyzer. Precision improved when analyzing with Global Southern in the analysis software and when excluding the 340-base fragment of GS-500. Although an issue for capillary electrophoresis, it appears not to be an issue on other instrument platforms. For example, no adverse effects on precision have been reported on the slab gel-based ABI Prism® 377 DNA Sequencer (PE Applied Biosystems, Foster City, California), where both the 250- and 340-base fragments are included in the Local Southern sizing algorithm (0.01-0.09 base standard deviation, Lazaruk et al. 1998; PE Applied Biosystems 1997). Even with sizing optimization on the 310 Genetic Analyzer, attempts should be made to control large ambient temperature fluctuations. Global Southern may allow for fluctuations that were unacceptable with Local Southern, but Global Southern sizing does not completely prevent temperature-shifted size deviations. Nonetheless, choosing Global Southern and excluding the 340-base internal fragment should improve precision and reduce off-ladder allele calls when ambient temperature fluctuations compromise precision. This method should also be considered for other STR multiplexes analyzed with GS-500 on the 310 Genetic Analyzer.
Elder, J. K. and Southern, E. M. Computer-aided analysis of one-dimensional restriction fragment gels. In: Nucleic Acid and Protein Sequence Analysis: A Practical Approach. M. J. Bishop and C. J. Rawlings, eds. Oxford IRL Press, 1987, Chapter 7, pp.165-172.
Elder, J. K. and Southern, E. M. Measurement of DNA length by gel electrophoresis II: Comparison of methods for relating mobility to fragment length, Analytical Biochemistry (1983) 128:227-231.
LaFountain, M. J., Schwartz, M. B., Svete, P. A., Walkinshaw, M. A., and Buel, E. TWGDAM validation of the AmpFlSTR® Profiler Plus™ and AmpFlSTR COfiler™ STR multiplex systems using capillary electrophoresis, Journal of Forensic Sciences (2001) 46(5):1191-1198.
Lazaruk, K., Walsh, S. P., Oaks, F., Gilbert, D., Rosenblum, B. B., Menchen, S., Scheibler, D., Wenz, M. H., Holt, C., and Wallin, J. Genotyping of forensic short tandem repeat (STR) systems based on sizing precision in a capillary electrophoresis instrument, Electrophoresis (1998) 19:86-93.
Moretti, T. R., Baumstark, A. L., Defenbaugh, D. A., Keys, K. M., Brown, A. L., and Budowle, B. Validation of STR typing by capillary electrophoresis, Journal of Forensic Sciences (2001) 46(3):661-676.
Pawlowski, R. and Maciejewska, A. The forensic validation studies of Profiler Plus™ and allele frequencies of profiler loci in a Polish population. In: The Tenth International Symposium on Human Identification. Promega Corporation, Orlando, Florida, 1999.
PE Applied Biosystems. GeneScan® Reference Guide. Foster City, California, 2000.
PE Applied Biosystems. GeneScan® Analysis Software User's Manual. Foster City, California, 1998.
PE Applied Biosystems. AmpFlSTR® Profiler Plus™ User's Manual, Version A. Foster City, California, 1997.
Rosenblum, B. B., Oaks, F., Menchen, S., and Johnson, B. Improved single-stranded DNA sizing accuracy in capillary electrophoresis, Nucleic Acids Research (1997) 25(19):3925-3929.