Coreexome 24

Genotype data were available from a previous study using microarray hybridization (CoreExome-24, version 1.1, Illumina, San Diego, CA, USA) [19 (link)]. The genomic positions of 144 genes, encoding HATs (12) and HDACs (18) and members of histone acetylation (59) and deacetylation (55) complexes, were identified, considering 1000 base pairs downstream and upstream from the sequence for the longest transcript, according to the human genome version GRC37/hg19 (NCBI gene) [20 (link)] (Table S1). A total of 2486 SNPs located in these gene sequences were extracted from DNA microarray data of EPF samples.
SNPs associated with EPF were genotyped in the SPF samples using the iPLEX platform of the MassARRAY system (Agena Bioscience, San Diego, CA, USA). The primer sequences are available in Table S2. MassARRAY Typer software (v4.0) (Agena Bioscience, San Diego, CA, USA) with standard settings was used to call the genotypes. The genotype distribution followed the Hardy–Weinberg equilibrium in patients, controls, and the set of participants (data not shown), except for rs13339618 (p_controls = 0.049; p_{all participants} = 0.005).

A total of 3937 Aotearoa NZ Polynesian and European individuals were genotyped in two batches (batch 1 with 291 and batch 2 with 3646 individuals) on the Illumina Infinium CoreExome-24 platform. (2997 used in this study were a subset.) From the data generated, log R ratios (LRRs) and B allele frequencies (BAFs) were extracted from the autosomes and used to identify CNV. Samples with a genotyping call rate of <98% and SNVs with a call rate of <95% were not utilized in the subsequent analysis (58 (link)). Principal component (PC) analysis was applied to the LRR values, and the first six PCs were visualized in scatter plots to evaluate the possibility of batch effects. Differential effects were not identified when samples were classified according to batch (Supplementary Material, Fig. S7) or ancestral group (Supplementary Material, Fig. S8). The Han Chinese controls of 1962 individuals were genotyped using the Affymetrix axiom genome-wide Chinese Han Beijiing array. Samples with a genotyping call rate of <95% and SNVs with a call rate of <95% were excluded. A total of 1848 Han Chinese samples were used in the subsequent analyses. Ninety-one Aotearoa NZ Polynesian individuals were sequenced to high coverage on a HiSeqX machine using TruSeq Nano libraries to allow validation of identified CNVRs.

DNA extracted from collected saliva samples was processed with an Illumina CoreExome24 array including 555,356 built-in markers that we further enriched with 4,000 custom markers (supplementary table S25, Supplementary Material online). A full description of DNA preparation and processing, including all undertaken quality control filtering steps, is provided in the supplementary text, , Supplementary Material online. The resulting final dataset, referred to as the ADME dataset, is composed of 349 genotyped DNA samples for 550,416 markers. The same markers were also successfully genotyped in an in-house collection of DNA samples from 119 Senegalese Mandenka, referred to as the SEN dataset (supplementary text, Supplementary Material online). To check the quality of the data and control for population structure, we ran principal components analyses (PCA) and ADMIXTURE (Alexander et al. 2009 (link)) on our dataset of five populations, as well as an additional PCA with 19 samples of populations from the 1,000 Genomes (1KG) project dataset (1000 Genomes Project Consortium, 2015 (link)) as a reference (fig. 1B, supplementary text, and Supplementary figs. 20 and 21, Supplementary Material online).