Genotyping and Association Analysis

Genotyping was performed using Illumina Human610-Quad v1.0 BeadChip array (Illumina, San Diego, CA, USA).^{[8 (link)]} The overall genotyping rate was 95.16%. After excluding rare (minor allele frequencies <5%) and low quality (call rate <90% and deviation from Hardy-Weinberg equilibrium P < 1e-3) variants, there are 533,687 remaining SNPs for the linear regression analysis. Using an additive genetic model, each SNP was tested for association to α-SMA expression and hepatic collagen content, respectively. The phenotypes were normalized with log base 10 transformation. Age, gender, body mass index, and the first two genetic principal components were adjusted as covariates for the association. SNPs with P value less than 1e-4 were considered as candidate loci for the following analysis. The quality control and association test was performed using the package PLINK 1.07 (http://pngu.mgh.harvard.edu/purcell/plink/).[^{12 (link)}] The regional plots were generated using the package LocusZoom (http://csg.sph.umich.edu/locuszoom/).^{[13 (link)]}