HLA row and benchmark call sets of HG002

Dataset OverviewThis dataset contains the variant calling data supporting the findings of the study "Long-Read Sequencing and the T2T-CHM13 Reference Improve HLA Variant Calling Accuracy". It provides a comprehensive evaluation of HLA variant calling performance using long-read/short-read sequencing technologies and CHM13/GRCh38 reference genomes. File Structure Description 01_Raw_Variant_Call_Sets: This folder contains the raw VCF files for the HLA region. We generated these files using HG002 sequencing data (35X coverage) from both PacBio HiFi and Illumina platforms. We aligned the reads to the CHM13, GRCh38, and hs38DH reference genomes. For alignment, we used Sentieon BWA for the Illumina data and pbmm2 for the PacBio HiFi data. After alignment, we called variants using both Sentieon DNAscope and DeepVariant. Please note that these raw VCF files have not undergone any normalization or filtering. 02_Benchmark_Call_Sets: This folder contains processed VCF files that we used directly for benchmarking and stratification analysis. To create these, we first extracted variant coordinates from the standard truth sets to make high-confidence BED files. We used these BED files to select high-confidence variants from the raw gVCF files, and the resulting VCF files were used for the subsequent stratified benchmarking.