GWAS and pathway analyses
Genotyping of the inbred lines in the panel was done via Genotype by Sequencing (GBS) according to Elshire et al. (2011). Briefly, SNPs were extracted from raw GBS data using the Java pipeline for GBS Bioinformatics (Glaubitz et al. 2014). SNPs were aligned against the B73 reference genome, version 4. The GBS marker dataset was imputed and filtered by removing SNPs with a minor allele frequency (MAF) < 6% and any SNP with > 2 alleles, resulting in a total of 1,105,817 SNPs. Useful GBS data was successfully obtained from 281 of the inbred lines, which were used in the GWAS. A subset consisting of 8,000 SNPs with a low missing data rate (< 7.5%) and a balanced allele frequency (MAF > 40%) was extracted for calculation of population sub-structure (Q matrix) and linkage disequilibrium, and 148,000 unimputed SNPs were used to calculate the K matrix. K, Q, and LD were calculated in the same manner reported in Warburton et al., (2015). The software package TASSEL 3.0 (Bradbury et al., 2007) was used to perform the GWAS using the BLUE values of the 7-day and 14-day FAW ratings within and across years. The General Linear Model (GLM) was run, as well as the Mixed Linear Model (MLM; Yu et al., 2006) using three subpopulations (Q matrix) and the K matrix. To correct for multiple comparisons, an adjusted Bonferroni-corrected threshold (i.e., P ≤ 1/N, where N is total number of genome-wide SNPs) was used for declaring the significance of GWAS associations (Benjamini and Hochberg 1995).
The Pathway Association Studies Tool (PAST; Thrash et al., 2020a) was used to perform a metabolic pathway analysis, as was first described in Tang et al. (2015). Linkage disequilibrium values between each marker and the 50 closest up- and downstream SNPs, and the SNP-trait association values for significance (p), correlation (R2 or the proportion of the phenotypic variation accounted for), and allele effect as calculated by TASSEL were used by the PAST program to calculate the running enrichment score for all annotated maize pathways. A running enrichment score measures the probability that each pathway is associated with FAW resistance. The PAST program assigned each SNP to an annotated gene, based on user defined linkage disequilibrium and physical distance values ofr2 > 0.8 and + 1 Kb, respectively. Each gene was assigned to a pathway using the gene annotation files in MaizeCyc (Monaco et al. 2013). Only pathways with five or more mapped genes (298 pathways) were considered in the analysis. More detail of the pathway analysis can be found in Warburton et al., (2017) and Li et al., (2019).