*3.5. GWAS for Bran WRC*

Among models tested, the best fit for the WRC trait was obtained using a GLM model with bran yield and friability as covariates (Figure 5). Unlike the friability trait, GWAS analysis did not reveal any makers that were significantly associated with WRC after Bonferroni correction (Figure 6). These results could be because the WRC data were not normally distributed or the phenotypic measurements recorded for WRC in this study are significantly influenced by nongenetic factors as noted by the low value of marker-based heritability. There were two markers located on the chromosome 4A that stood out from among the other SNPs: IWA4867 (562,407,344 bp) and IWA4698 (562,453,542 bp). Despite being not significant, these markers were examined further to determine if there may be any logical link to bran WRC. A majority of genotypes had these alleles for increased WRC: 192 genotypes for marker IWA4867 (TT genotype) and 194 for marker IWA4698 (TT genotype; Table 1; Supplementary Table S1). These genotypes were randomly distributed on a PCA biplot generated using the marker data, suggesting the absence of groupings among lines carrying the favorable alleles (Supplementary Figure S2).

As indicated, most genotypes carried the tentative alleles for increased bran WRC. Unfortunately, higher WRC of bran is unfavorable, because it is associated with production of bread with low loaf volume [9]. While the vast majority of breeding e fforts have been focused on improving the performance of the refined flour, limited attention was dedicated to bran quality. Therefore, more e fforts are needed to select lines for whole grain baking. Perhaps this can be done by selecting lines that do not have these alleles.

**Figure 5.** Quantile-quantile plots for water retention capacity trait: (**a**) GLM model with no covariates; (**b**) GLM model with bran yield (BY) as covariate; (**c**) GLM model with bran friability (BF) as covariate; (**d**) GLM model with BY and BF as covariates; (**e**) MLM model with kinship and no covariates; (**f**) MLM model with kinship and BY as covariate; (**g**) MLM model with kinship and BF as covariate; (**h**) MLM model with kinship and BY and BF as covariates.

**Figure 6.** Manhattan plot showing trait associated SNPs for water retention capacity; Un, unanchored scaffold; red line, Bonferroni correction (adjusted *p* < 0.05) cutoff; blue line, notable SNPs above the background.

### *3.6. Candidate Genomic Regions for Bran Friability and WRC*

The two markers that were significantly associated with bran friability, BS00000020\_51 (5D: 3,609,894 bp) and Excalibur\_c49805\_63 (5D: 1,614,602 bp), were located in genes *TraesCS5D02G004300* (5D: 3,609,672–3,610,121 bp) and TraesCS5D02G001200 (5D: 1,609,486–1,614,664 bp). *TraesCS5D02G001200* encodes for sucrose membrane transfer proteins.As a part of cell membranes, these proteins could play an additional role in the friability of bran during milling. The other gene, *TraesCS5D02G004300*, was annotated as *Pinb-D1b* at Ensembl Plants [63] and as *Puroindoline-b* in the IWGSC functional annotations [62]. A BLASTN search with Pinb-D1 forward primer sequence (obtained from Graingenes; [66]) as query against the IWGSC RefSeq v1.0 wheat genome matched exactly to the first 20 bp of this gene. Further, the correlation coefficient between the SNP marker associated with bran friability (BS00000020\_51) and *Pinb-D1b* marker (r = 0.75; *p* < 0.001) and *Pinb-Wild* (r = −0.70; *p* < 0.001) was relatively stronger compared to *Pina-D1* (r = 0.35; *p* = 0.02). This result is probably because the distance between *Pinb-D1* forward primer start position and the friability-associated SNP marker (BS00000020\_51) is 222 bp. Overall, these observations indicate that the SNP marker associated with bran friability (BS00000020\_51; 5D: 3 609 894 bp) is in linkage disequilibrium and potentially tagged to the Pinb-D1 marker, and is a functional variant for *Pinb* locus. This also suggests that *Pinb-D1* locus may be involved in regulating bran friability.

*Pinb-D1* genes encode for puroindoline proteins, which make up a complex multicomponent complex called fraibilin [67]. Friabilin is involved in the association between starch and protein in the wheat kernel and is responsible for soft endosperm texture [68]. The results from this study sugges<sup>t</sup> that, in addition to controlling endosperm texture, the puroindoline genes may also affect bran texture.

The two SNP markers that were tentatively associated withWRC, IWA4867 (4A: 562,407,344 bp) and IWA4698 (4A: 562,453,542 bp), were located close to *TraesCS4A02G251100* (562,408,612–562,414,155 bp) and *TraesCS4A02G251300* (562,457,963–562,459,338 bp), respectively. The *TraesCS4A02G251300* gene did not have a characterized function in wheat. However, it did have high similarity to the *Zm00001d033902\_T001* gene present in maize (%ID—89.5; E-value: 1.6 × <sup>10</sup>−63), which was reported to encode for cysteine-rich secretory proteins. Interestingly, in wheat, nongluten cysteine-rich proteins are harmful for breadmaking because they interfere with gluten macropolymerization [69].
