We demonstrate that wgaDNA, made from 3.2 mm disks of DBSS that have been stored at -24°C for more than 20 years is well suited for reliable genotyping of 610,000 SNPs, with call-rates comparable to those obtained using gDNA. The accuracy of genotype calls using wgaDNA from stored DBSS has been of some concern. The issue has been addressed several times, using both low and medium throughput genotyping platforms, and overall with good success [11, 12, 14, 15, 16, 17, 18]. In this study we took the usage of DBSS one step further by conducting GWS. Moreover the accuracy of genotype calls from wgaDNA was assessed by comparing the results with results from high-quality reference gDNA from the same individuals. Initially, we tested two commercial DNA-extraction procedures, three WGA procedures, the effect of number of 3.2 mm disks used, and the effect of protein extraction prior to the gDNA extraction. The efficiency and reliability of the GWS were highly dependent on the employed DNA-extraction and WGA method. Interestingly, call- and conflict-rates were inversely related; indicating that genome scan of wgaDNA is highly reliable when the call-rates are close to 100%. However because only few samples were available to calculate the correlation coefficient, we cannot clearly define a cutoff for the call-rate that would ensure reliable genotyping. In general, the OmniPlex method performed more constantly than the MDA method, producing fairly the same call-and conflict-rates independently of the other variables tested. Of the two OmniPlex based kits the GPlex4 kit performed the best, showing high call-rates and low error-rates. The MDA method performed excellent using the ENA extraction kit and poorly when using the QIA extraction kit. In general, it appeared unimportant whether one or three DBSS disks were used for extraction. This was surprising since the amount of input gDNA for the WGA reactions is supposed to be critical, and in our set-up it was often below the lower limit of 10 ng that is recommended by the manufacturer. Moreover, the preceding protein extraction ofthe disks did not impair the genotyping of the produced wgaDNA, which is in accordance with similar observations from our laboratory .
Because the investigation focuses ondifferent combinations of wgaDNA preparation, it suffers from the weakness that the number of samples in each group is limited. In addition, only samples from the DNSB were used.
The combination of the ENA DNA-extraction with either the REPLI-g or the Gplex4 WGA kit were selected to see if the procedures were robust enough for GWA studies employing numerous samples. Both set-ups produced wgaDNA from 16 DBSS stored for 15 to 25 years that performed well with constant high call-rates. Corresponding reference gDNA samples were not available. Notably, when calling genotypes of wgaDNA preparations with the BeadStudio software, albeit clusters were nice and tight for some loci they did not fit well into the standard cluster positions. This is because the BeadStudio software calls the genotypes of a given locus by comparing the observed values with the expected values, defined by the Human610-Quadv1B Cluster file, which is based on gDNA samples [22, 23]. In such cases, data fit and call-rates can be improved by adjusting the cluster positions to match the observed data . Cluster files tailored for the OmniPlex and MDA method were hence created from the samples available, and the call-rates were significantly improved for both wgaDNA preparations. They were in fact comparable to call-rates obtained using high-quality gDNA, indicating that the approach is robust.
Eighteen WGA reactions, each producing ~5 μg of wgaDNA, can be made per ENA DNA extraction. As the Illumina Infinium HD Human610-Quad BeadChip uses 240 ng of wgaDNA, one WGA reaction is enough to run 20 chips. Thus one to three 3.2 mm disks from a DBSS are sufficient to make repeated GWS as well as fine-mapping genotyping, if required. We have briefly tested the performance of the two wgaDNA preparations on the Affymetrix platform and found that wgaDNA produced by the OmniPlex method was unsuitable, whereas wgaDNA produced by the MDA method gave results comparable to those obtained by the Illumina platform. In addition to being used for GWS, DBSS can also be used for multiplex protein measurements , quantitative RNA micro arrays detecting up to 3000 genes , and quantitative DNA methylation analysis .