Initial replication analysis was of cohorts used in our previous work [12], with ethnic outliers removed as described. Obesity case-control cohorts from France were as previously published [17]: Phenotypes and genotyping data (Illumina Human CNV370-duo arrays) passing quality control were available for 649 obese children with a body mass index (BMI) ≥97th percentile corrected for gender and age; 557 non-obese controls (BMI ≤90th percentile); 705 obese adults (BMI ≥40 kg.m−2) and 843 non-obese controls (<25 kg.m−2). Data for 141 severely obese French patients undergoing elective bariatric weight-loss surgery were as previously described [12]. For The Northern Finland Birth Cohort 1966 (NFBC1966) prospective birth cohort [18], phenotypic data and genotyping data (Illumina Human CNV370-duo arrays) passing quality control was available for 5,216 subjects aged 31 years at the time of phenotyping. For further replication of the 220 kb deletion on chromosome 16p11.2, genotyping data was available for other previously-described population cohorts as follows: the CoLaus prospective population cohort [44] –5,612 white individuals aged 35–75 years randomly selected from the general population in Lausanne, Switzerland; the EGCUT BioBank [45] –2,666 individuals randomly
