Background/Purpose: In Genome-Wide Association Studies (GWAS), the performance of the genotyping algorithm is crucial to identify new SNPs associated to disease risk. Recently, new methods have been developed to exploit SNP-oriented microarrays in order to genotype Copy Number Variations (CNVs), albeit with a reduced performance compared to SNPs. In this study we present CNstream2, a method for both SNP and CNV genotyping that achieves a superior accuracy compared to other established methods. In addition, we have identified several new CNV loci that are in high linkage disequilibrium (LD) with SNPs previously associated to rheumatic diseases.

Methods: CNstream2 is a substantially improved version of our previous CNstream software which achieves a superior accuracy in both SNP and CNV genotyping compared to other well-established methods (i.e. GenoSNP/GenCall for SNP genotyping and PennCNV/QuantiSNP for CNV genotyping). All these improvements have been assessed in different Illumina platforms using public microarray data from HapMap samples. SNP genotypes from the 1000 Genomes Project (1KGP) and CNV genotypes from recent studies using CNV-oriented technologies have been used as golden standards for performance comparisons. In order to show the power of CNstream2, we performed a correlation analysis between SNPs associated with rheumatic diseases reported by the GWAS catalog and the CNVs identified by CNstream2 over the same HapMap samples using HumanOmni1 data. All the CNV loci that obtained an r2>0.7 (N=16) were reported in this study.

Results: CNstream2 obtained a high performance both on SNP and CNV genotyping. When assessing SNP genotyping accuracy, CNstream2 obtained an average gain of 0.20% with respect to GenoSNP and GenCall (representing a gain of 1,000-2000 genotyped SNPs per GWAS). On the other hand, when comparing CNV calls obtained by CNstream2, PennCNV and QuantiSNP within previously characterized CNV loci, CNstream2 exceeded by an average of 20% the number of correctly captured loci compared to its competitors. The LD analysis between SNPs associated to rheumatic diseases and CNVs detected by CNstream2 revealed 16 highly correlated SNP-CNV pairs. From these, 11 pairs were located in the HLA region and were associated to several rheumatic diseases. Given the strong correlation of the CNV with the disease risk SNP, additional functional studies exploring its relevance are warranted. Outside the HLA region, 5 new CNVs from loci associated with rheumatological diseases were identified. From these, a previously unidentified intronic deletion in Rheumatoid Arthritis risk gene PADI4, showed a strong association. 

Conclusion: After an exhaustive evaluation of CNstream2 performance we can conclude that this new software tool provides an unprecedented accuracy both in SNP and CNV genotyping. Using CNStream2 on publicly available data, we have identified new CNVs on loci previously associated to rheumatological diseases which could likely explain the observed disease risk association.

---

« Back to 2012 ACR/ARHP Annual Meeting

ACR Meeting Abstracts - https://acrabstracts.org/abstract/cnstream2-improved-snp-and-cnv-genotyping-reveals-new-loci-associated-with-rheumatic-diseases/