Asset Tag: Rare disease

Introduction: There are many clinically important genes in “dark” regions of the human genome. These regions are characterized as dark due to a paucity of NGS coverage as a result of short-read sequencing or mapping difficulties. Low NGS sequencing yield can arise in these regions due to the presence of various repeat elements or biased base composition while inaccurate mapping is attributable to segmental duplications. Long-read sequencing coupled with an optimized, robust enrichment method has the potential to illuminate these dark regions.

Materials and Methods: Using PacBio highly accurate long-read (HiFi) Sequencing, coupled with a long-PCR targeted enrichment method, we investigated two important dark region genes that are challenging to accurately type with short-read sequencing due to associated pseudogenes: CYP21A2, responsible for congenital adrenal hyperplasia, and GBA, responsible for Gaucher disease. For each gene, our aim was to cover regions of pathogenic mutations in a single contiguous sequence or set of sequences that can be assayed in a single reaction. CYP21A2 and an associated pseudogene CYP21A1P were co-amplified in a single long-range PCR reaction generating a 10.2 kb and 8.9 kb amplicon, respectively. Similarly, GBA and an associated pseudogene GBAP1 were co-amplified in a single long-range PCR reaction generating a 12.6 kb and 16.0 kb amplicon, respectively. Seven Coriell samples for the CYP21A2 target region and 13 Coriell samples for the GBA target region containing known pathogenic mutations were studied in replicate. SMRTbell libraries were generated from pooled amplicons for each target gene and sequenced on a PacBio Sequel II System. Accounting for replicates, each library contained a multiplex of 24 samples. A new PacBio sequence clustering algorithm, pbAA, designed for rapid analysis of HiFi reads from amplicons was used in variant typing.

Results: All pathogenic CYP21A2 and GBA variants were accurately called in the test samples. These variants included whole-gene deletions, gene duplication, gene fusions, and recombinant exons. Additionally, phasing of complex heterozygotes was achieved.

Conclusion: We demonstrate that long-read HiFi Sequencing provides new opportunities for sequencing clinically relevant but previously dark regions of the human genome that are underrepresented in short-read sequencing. Accurate long reads provide important phasing information, identify structural variations, and avoid potential confusion with pseudogenes. SMRT Sequencing of these regions enables a better understanding of the relationship between genetic factors and personal health and has the potential to ultimately help guide health-related decisions.

Many genetic diseases are mapped to structurally complex loci. These regions contain highly similar paralogous alleles (>99% identity) that span kilobases within the human genome. Comprehensive screening for pathogenic variants amongst paralogous sequences is incomplete and labor intensive using short-reads or optical mapping. In contrast, long-range targeted amplification and PacBio HiFi sequencing fully and directly resolves and phases a wide range of pathogenic variants without assembly or inference. To capitalize on the accuracy of HiFi amplicon data we designed a new amplicon analysis tool, pbAA. pbAA uses a new sequence clustering algorithm to rapidly deconvolve (separate) a mixture of haplotypes, enabling precise diplotyping, and disease allele classification. In this experiment, we analyzed two sets of gene-pseudogene systems, GBA and CYP, that are the second and eighth most common carrier disease alleles, respectively. Samples tested were selected from the Coriell catalog known to have pathogenic variants troublesome to test for with standard short-read assays. Co-amplified long-range PCR amplicons were generated for GBA (12kb)/GBAP1 (15kb), responsible for Gaucher disease, as well as CYP21A2 (10kb)/CYP21A1P (8kb), responsible for congenital adrenal hyperplasia. We obtained 7 samples to test the CYP21A2 region and 13 separate samples for GBA.  HiFi reads were then generated from the amplicon libraries on both Sequel and Sequel II Systems, with replicated samples, to achieve a 24-sample multiplex for each target. Consensus amplicons were produced using pbAA and variants were determined using minimap2 alignments along with a custom SQL database for characterizing and reporting results.  From these data we were able to accurately call all pathogenic variants in the test samples for all replicates, including whole-gene deletions, gene duplication, gene fusions, recombinant exons, and phased complex heterozygotes.  In one trio affected by adrenal hyperplasia, three large structural variants were correctly and independently attributed to the parents and proband, including a duplication of CP21A1P and a CYP21A1P-CYP21A2 gene fusion in the mother and a CYP21A2 deletion in the father. This experiment demonstrates how PacBio HiFi data, analyzed with pbAA, simplifies targeted disease allele identification.