Axiom hd

gDNA samples were genotyped using 600 K Affymetrix Axiom HD genotyping arrays containing 580,954 SNPs based on the Affymetrix GeneChip platform [29 (link)]. The genotypes were identified using Affymetrix Genotyping Console (Version 4.2, Affymetrix Inc., Santa Clara, CA, USA) and a custom cluster file developed from the 96 (48 deformed and 48 normal) samples. Following the case-control design, stringent QC procedures were performed for the genotype data using PLINK [56 (link)] (Version 1.07, http://zzz.bwh.harvard.edu/plink/). First, individuals with > 10% missing genotypes were excluded (n = 1). Second, out of the initial full set of 554,305 effective SNPs in this study, we discarded those with a call rate < 90% (n = 4505) and those having a minor allele frequency (MAF) < 0.05 in all birds (n = 120,261). The Hardy-Weinberg equilibrium (HWE) was not used to filter data in view of the small population.

To enumerate Campylobacter, serial ten-fold dilution series of weighed contents of the two caeca were separately prepared to 10^–7 in phosphate-buffered saline and 100 μL of each dilution plated to modified charcoal deoxycholate (mCCDA) agar supplemented with cefoperazone (32 mg/L) and amphotericin B (10 mg/L; Oxoid), followed by incubation for 48 h under microaerophilic conditions (5% O₂, 5% CO₂, and 90% N₂) at 41 °C. Dilutions were plated in duplicate and colonies with morphology typical of Campylobacter were counted and expressed as CFU/g. The theoretical limit of detection by the method used was 100 CFU/g. In instances where no colonies were observed after direct plating, a Campylobacter load equal to the theoretical limit of detection was assumed, as enrichment to confirm the absence of Campylobacter in caecal contents was not performed.
All the birds were genotyped with a proprietary 50K high-density genome-wide SNP array and then imputed using AlphaImpute^{73 (link),74 (link)} to the 600K SNP Affymetrix Axiom HD array^{75 (link)} based on parent, grand-parent and great-grand-parent 600K SNP Affymetrix data. Of 3000 birds sampled, genotypes for 2718 birds were successfully imputed. Imputation failures likely reflect a lack of compatibility between the pedigree information and the genotypic data.