Comparison of Multiple Methods for Determination of FCGR3A/B Genomic Copy Numbers in HapMap Asian Populations with Two Public Databases

Abstract

Low FCGR3 copy numbers (CNs) has been associated with susceptibility to several systemic autoimmune diseases. However, inconsistent associations were reported and errors caused by shaky methods were suggested to be the major causes. In large scale case control association studies, robust copy number determination method is thus warranted, which was the main focus of the current study. In the present study, FCGR3 CNs of 90 HapMap Asians were firstly checked using four assays including paralog ratio test combined with restriction enzyme digest variant ratio (PRT-REDVR), real-time quantitative (qPCR) using TaqMan assay, real-time qPCR using SYBR Green dye and short tenden repeat (STR). To improve the comparison precision reproductively, the results were compared with those from recently released sequencing data from 1000 genomes project as well as whole-genome tiling BAC array data. The tendencies of inconsistent samples by these methods were also characterized. Refined in-home TaqMan qPCR assay showed the highest correlation with array-CGH results (r = 0.726, p < 0.001) and the highest concordant rate with 1000 genome sequencing data (FCGR3A 91.76%, FCGR3B 85.88%, and FCGR3 81.18%). For samples with copy number variations, comprehensive analysis of multiple methods was required in order to improve detection accuracy. All these method were prone to detect copy number to be higher than that from direct sequencing. All the four PCR based CN determination methods (qPCR using TaqMan probes or SYBR Green, PRT, STR) were prone to higher estimation errors and thus may lead to artificial associations in large-scale case-control association studies. But different to previous reports, we observed that properly refined TaqMan qPCR assay was not inferior to or even more accurate than PRT when using sequencing data as the reference.

AMYCNE: Confident copy number assessment using whole genome sequencing data

“…Copy number measurements of FCGR3A and FCGR3B was available for 164 individuals from the 1000 genomes cohort [ 14 , 16 ]. The low coverage whole genome data available through the 1000 genomes consortium from the 164 individuals was analyzed using AMYCNE and CNVnator and the results were compared to the truth set presented in [ 16 ], using the same procedure as presented for the AMY1 locus. DCNE of AMY1 was not applied to this test since it is designed specifically for genotyping of AMY1 .…”

GeneToCN: an alignment-free method for gene copy number estimation directly from next-generation sequencing reads

“…Copy numbers of FCGR3A and FCGR3B genes were estimated for 164 individuals using the low coverage whole genome sequencing data from 1000 Genome project 33 . The copy numbers for the same individuals have previously been estimated using AMYCNE and CNVnator 13 and compared to the copy numbers that were determined using multiple different methods and therefore presumed to be the correct copy numbers 34 . Using the same truth set for comparison, the concordance of the copy number estimations for FCGR3A was 0.74 (0.71 and 0.5 from AMYCNE and CNVnator, respectively), while for FCGR3B, it was 0.63 (0.85 from both AMYCNE and CNVnator).…”

Noise-robust assessment of SNP array based CNV calls through local noise estimation of log R ratios

“…As BAF profiles seem often not informative in the regions where calls are made incorrectly, it might be interesting to see if additional belief thresholds for extreme values can be identified. Another weakness of this study is the use of qPCR as reference method, because qPCR is also not free of errors, especially in case of segmental duplications and complex structural rearrangements (Fernandez-Jimenez et al, 2011;Qi et al, 2016). However, this noise estimation procedure can also be extended to other CNV calling platforms such as sequencing based approaches.…”