Kun:LabNotes/CpgSeq/2010-7-21

From ZhangLabWiki
Jump to navigation Jump to search

mQTL analysis on the Breast Cancer data[edit]

  • Data used:
    • Methylation data: 30 samples captured with the Cpg97k probe set provided by Dinh (0_bc_matrix.txt);
    • SNP data: 16 Illumina 1M SNP data set generated by Beckman;
  • Procedure:
  1. Filtering of DNA methylation data to select for variable Cpg sites: I calculated the STD for each site, and remove those with STD<0.1; I also removed all sites that have >20% missing values. Script.
  2. Filtering of SNP data: I removed SNPs with MAF<0.2 and those has >25% missing genotypes. Script;
  3. For each CpG site, take all SNPs within 1Mb downstream and upstream, perform association test with PLINK;
  4. Estimate p-values based on permutation;
  5. Select CpG-SNP pairs at FDR<0.1. Script.
  • Results: 34,051 pairs (26,481 SNPs; 2,762 CpG sites (15.8%))