Wheat breeding for FHB resistance is a complex undertaking that relies heavily on the genetic variation among candidate germplasm. The quantitative nature of FHB resistance has made the task even more difficult. More than 500 QTL for FHB resistance have been detected in diverse wheat accessions worldwide, but characterization of the genes underlying these QTL has been limited. To improve the efficacy of the QTL in FHB resistance and to characterize the underlying genes, we created a wheat practical haplotype graph pangenome database using the Practical Haplotype Graph (PHG) tool. Whole exome-capture sequencing data from 95 spring wheat genotypes used in ND, SD, MN, and MT wheat breeding programs were used to build the PHG database. Based on the high confidence gene models from IWGSC Refseq v1.1, a set of 94,229 reference genome intervals were used to infer haplotypes from the wheat lines. In this ongoing study, we estimated the accuracy of PHG -imputed genotype calls in the wheat lines that were genotyped by exome capture, or whole-genome skim-sequencing approach. The PHG imputation accuracy varied between 97 - 98% for the exome capture data when compared against high confidence exome capture data concordant with 90K array sites. The accuracy of imputed skim-sequenced data with the PHG database ranged between 75-90%. The imputation accuracy for the sequencing data used in this study using other imputation methods was generally higher than the PHG but decreased as minor allele frequency increased. Further development of the PHG database is needed to evaluate its potential for large-scale imputation.