Assessing the utility of whole-genome amplified serum DNA for array-based high throughput genotyping

Abstract

Background: Whole genome amplification (WGA) offers new possibilities for genome-wide association studies where limited DNA samples have been collected. This study provides a realistic and high-precision assessment of WGA DNA genotyping performance from 20-year old archived serum samples using the Affymetrix Genome-Wide Human SNP Array 6.0 (SNP6.0) platform.

Results: Whole-genome amplified (WGA) DNA samples from 45 archived serum replicates and 5 fresh sera paired with non-amplified genomic DNA were genotyped in duplicate. All genotyped samples passed the imposed QC thresholds for quantity and quality. In general, WGA serum DNA samples produced low call rates (45.00 +/- 2.69%), although reproducibility for successfully called markers was favorable (concordance = 95.61 +/- 4.39%). Heterozygote dropouts explained the majority (>85% in technical replicates, 50% in paired genomic/serum samples) of discordant results. Genotyping performance on WGA serum DNA samples was improved by implementation of Corrected Robust Linear Model with Maximum Likelihood Classification (CRLMM) algorithm but at the loss of many samples which failed to pass its quality threshold. Poor genotype clustering was evident in the samples that failed the CRLMM confidence threshold.

Conclusions: We conclude that while it is possible to extract genomic DNA and subsequently perform whole-genome amplification from archived serum samples, WGA serum DNA did not perform well and appeared unsuitable for high-resolution genotyping on these arrays.

Figure 1

Relative call rates between archived samples and controls. Average genotype call rates from Birdseedv2.0 analysis at a default confidence threshold of 0.1 showed relative performance of genotyped samples on SNP 6.0 platform.

Figure 2

Scattergram of call rates between serum technical replicates. Call rates between serum technical replicates were plotted against each other to assess consistency of performance of technical replicates on the SNP 6.0 genotyping platform.

Figure 3

Performance analysis of technical replicate samples.

Figure 4

Birdseedv2.0 vs CRLMM SNP 6.0 genotyping performance on WGA samples. Assessment of genotype calling performance of Birdseed v2.0 and CRLMM algorithms from WGA DNA from serum samples on SNP6.0 genotyping platform is shown by comparison of (a) call rates (b) concordance between technical replicates and (c) paired sample concordance.

Figure 5

Chip quality assessment of good performing sample vs bad performing sample on SNP 6.0 genotyping platform. Scattergram plots of probe intensity values for allele B (Theta B) against allele A (Theta B) were generated from arrays of a (a) non-amplified genomic DNA sample, (b) good performing WGA DNA from serum and (c) bad performing WGA DNA from serum. Each point in the plot represents an assayed marker on the SNP 6.0 platform. Dark areas in the plot represent space with high-density points whereas light areas represent space with low-density points.