IL-12 RB1 Genetic Variants Contribute to Human Susceptibility to Severe Acute Respiratory Syndrome Infection among Chinese

Tang, Fang; Liu, Wei; Zhang, Fang; Xin, Zhong-Tao; Wei, Mao-Ti; Zhang, Pan-He; Yang, Hong; Ly, Hinh; Cao, Wu-Chun

Genotype distributions of each SNPs were tested for the Hardy-Weinberg equilibrium by the Chi square test in three groups. The genotype frequencies of the SARS patients and the control subjects were compared, and odds ratios (OR) were calculated by assigning the reference value (1.0) to the homozygous genotype that was more frequent in the control subjects than in the SARS patients. The OR and P values were calculated by multivariate logistic regression analysis with age, sex , comorbidity presence and infection source entered as covariates in the model, given the relevance of these potential confounding variables in interpreting the study results. Data were analyzed using the SPSS software (version 10.0, SPSS Inc, Chicago, IL, USA). To correct for multiple testing, we used spectral decomposition (SpD) of matrices of pairwise linkage disequilibrium (LD) between SNPs, which adjusts for multiple testing while taking into account LD among the tested SNPs [21] . This method showed that P-values of 0.0227 and below can be considered as statistically significant after correction for multiple testing. The software HAPLOVIEW (version 2.05) was used to construct haplotypes from the genotype data. This program calculates association statistics for multilocus haplotypes in case-control data using the expectation-maximisation algorithm to estimate haplotype odds ratios across multiple categories, giving a likelihood ratio test of homogeneity.