Table S1 from An integrative machine learning approach to discovering multi-level molecular mechanisms of obesity using data from monozygotic twin pairs

Table of SNPs used in the analysis and the sources from which they were chosen To focus our analysis on the most important findings, we chose to include in the analyses SNPs associated with obesity and obesity related traits. An additional motivation for this choice is that, as we found in previous work (28), it is difficult to draw meaningful or actionable hypotheses from genes for which nothing is known. The main article used to choose the SNPs for this analysis was the GWAS meta-analysis of Locke et al (8). We also used SNPs retrieved from searches for BMI, liver disease, metabolic syndrome and diabetes from the NHGRI-EBI GWAS Catalog (104) as well as the SNPs from the paper of Turcot et al (105) on rare variants associated with BMI. The number in be-tween dollar signs links to the source from which each SNP was chosen, and is also included in the component diagrams.