Genechip mapping 5.0 k array

CMC count data was ingested and processed similar to the AMP-AD transcriptome data. An iterative normalization model was deployed to identify significant covariates and regress them from the expression data before scaling the data (Additional file 2: Table S2). Genotype data was profiled with Affymetrix GeneChip Mapping 5.0 K Array and a custom version of the Illumina Infinium CoreExome-24 v1.1 BeadChip (#WG-331-1111). Raw data was filtered to remove SNPs with: zero alternate alleles, MAF < 1%, genotyping rate < 0.95, Hardy-Weinberg p value < 1 × 10⁻⁶, and individuals with genotyping rate < 0.95. Imputation was performed using eagle, Minimac, and the HRC Reference Panel [18 ]. Imputed variant data was filtered for SNPs present in the LD reference panel using Plink (v1.9). CMC data was withheld from training gene weight models for the purpose of validating gene weights in an independent cohort, blinded from the training models. Expression values were imputed, and Kendall correlation values were calculated comparing imputed gene expression to the scaled, assayed expression values. Correlation test values were FDR corrected for the number of matched comparisons N = 6643.