SNPs selection

Polymorphisms in miR-17-92 cluster binding sites were selected using bioinformatics databases: miRNASNP (http://bioinfo.life.hust.edu.cn/miRNASNP2/), mirsnpscore (http://www.bigr.medisin.ntnu.no/mirsnpscore/), PolymiRTS (http://compbio.uthsc.edu/miRSNP/), and MirSNP (http://bioinfo.bjmu.edu.cn/mirsnp/search/). In the miRNASNP database, putative miRNA target binding sites were predicted by TargetScan and miRmap tools [17]. The mirsnpscore online tool predicted the effects of SNPs in miRNA binding sites and mapped these miRNA-related variants to interested SNPs in GWAS using linkage disequilibrium [25]. The PolymiRTS database identified the predicted binding sites of interested SNPs and the miRNA seed regions using the TargetScan algorithm [26]. The MirSNP database provides human SNPs that located at miRNA-mRNA binding sites, which were mainly predicted by miRanda algorithm [27]. The candidate miR-SNPs and miRNA target binding sites were obtained by searching the miR-17-92 cluster in above databases. We also searched published literatures to screen cancer-related target genes. The HapMap CHB database was used to select SNPs according to the criteria of a minor allele frequency (MAF) > 5% in the Chinese Han population. Finally, six SNPs (rs12594531, rs1366600, rs1804506, rs3741779, rs3763763, and rs8323) were selected as candidate SNPs within the binding sites of miR-17-92 cluster. The selected miR-SNPs and the predicted binding sites between miRNAs and target genes were presented in Supplementary Fig. S1 (Additional file 1).