Accuracy in copy number calling by qPCR and PRT: a matter of DNA

Abstract

The possible implication of copy number variation (CNV) in the genetic susceptibility to human disease needs to be assessed using robust methods that can be applied at a population scale. In this report, we analyze the performance of the two major techniques, quantitative PCR (qPCR) and paralog ratio test (PRT), and investigate the influence of input DNA amount and template integrity on the reliability of both methods. Analysis of three genes (PRELID1, SYNPO and DEFB4) in a large sample set showed that both methods are prone to false copy number assignments if sufficient attention is not paid to DNA concentration and quality. Accurate normalization of samples is essential for reproducible qPCR because it avoids the effect of differential amplification efficiencies between target and control assays, whereas PRT is generally more sensitive to template degradation due to the fact that longer amplicons are usually needed to optimize sensitivity and specificity of paralog sequence PCR. The use of normalized, high quality genomic DNA yields comparable results with both methods.

Figure 1. Schematic representation of the PRT assay for PRELID1.

Forward and FAM-labeled reverse primers amplify two 299 bp fragments from different genomic locations (PRELID1, the target gene on chromosome 5 and a paralog on chromosome 1). Bcl I digestion results in a 169 bp labeled fragment in the paralog amplicon, which can be distinguished from the undigested PRELID1 fragment by fluorescent capillary electrophoresis. Coordinates correspond to GRCh37/hg19 genome assembly.

Figure 2. Distribution of PRELID1 and SYNPO calculated copy numbers in normalized (black bars) and randomly diluted (gray bars) DNA samples using qPCR.

Samples falling into the light gray areas are predicted to have two copies of the gene.

Figure 3. Correlation between the amount of input DNA per qPCR reaction and the raw copy number values for PRELID1.

Figure 4. Influence of DNA integrity on PRELID1 copy number assignment by qPCR and PRT analysis.

Distribution of calculated copy number values (black bars) and frequency of predicted copy numbers (gray bars) in a) high quality and b) degraded DNA samples, and gel electrophoresis of a subset of each sample set. Lanes 1 and 2 correspond to DNA size markers (λ-HindIII and 123 bp ladder) and 3–10 to genomic DNA samples. The size of several marker bands is shown in base pairs. The distribution of predicted copy number values differ significantly between good quality and degraded samples in the PRT analysis (χ² = 38.34; p = 2·10⁻⁸) and between qPCR and PRT copy number assignments in degraded DNA samples (χ² = 33.96; p = 2·10⁻⁷).

Figure 5. Distribution of calculated copy number values for DEFB4 obtained with qPCR and PRT in 366 normalized, high quality DNA samples.

Figure 6. Correlation of calculated DEFB4 copy number values between replicates in qPCR (R = 0.8546) and PRT (R = 0.8193) techniques.

Figure 7. Bland-Altman plot showing deviation of DEFB4 calculated copy number values obtained by PRT, compared to qPCR.