Portal de Periódicos da CAPES

Long-Read Nanopore Sequencing Validated for Human Leukocyte Antigen Class I Typing in Routine Diagnostics

2020; Elsevier BV; Volume: 22; Issue: 7 Linguagem: Inglês

10.1016/j.jmoldx.2020.04.001

1943-7811

Benedict M. Matern, Timo I. Olieslagers, Mathijs Groeneweg, Burcu Duygu, Lotte Wieten, Marcel G.J. Tilanus, Christina E.M. Voorter,

Immune Cell Function and Interaction

Matching of human leukocyte antigen (HLA) gene polymorphisms by high-resolution DNA sequence analysis is the gold standard for determining compatibility between patient and donor for hematopoietic stem cell transplantation. Single-molecule sequencing (PacBio or MinION) is a newest (third) generation sequencing approach. MinION is a nanopore sequencing platform, which provides long targeted DNA sequences. The long reads provide unambiguous phasing, but the initial high error profile prevented its use in high-impact applications, such as HLA typing for HLA matching of donor and recipient in the transplantation setting. Ongoing developments on instrumentation and basecalling software have improved the per-base accuracy of 1D2 nanopore reads tremendously. In the current study, two validation panels of samples covering 70 of the 71 known HLA class I allele groups were used to compare third field sequences obtained by MinION, with Sanger sequence–based typing showing a 100% concordance between both data sets. In addition, the first validation panel was used to set the acceptance criteria for the use of MinION in a routine setting. The acceptance criteria were subsequently confirmed with the second validation panel. In summary, the present study describes validation and implementation of nanopore sequencing HLA class I typing method and illustrates that nanopore sequencing technology has advanced to a point where it can be used in routine diagnostics with high accuracy. Matching of human leukocyte antigen (HLA) gene polymorphisms by high-resolution DNA sequence analysis is the gold standard for determining compatibility between patient and donor for hematopoietic stem cell transplantation. Single-molecule sequencing (PacBio or MinION) is a newest (third) generation sequencing approach. MinION is a nanopore sequencing platform, which provides long targeted DNA sequences. The long reads provide unambiguous phasing, but the initial high error profile prevented its use in high-impact applications, such as HLA typing for HLA matching of donor and recipient in the transplantation setting. Ongoing developments on instrumentation and basecalling software have improved the per-base accuracy of 1D2 nanopore reads tremendously. In the current study, two validation panels of samples covering 70 of the 71 known HLA class I allele groups were used to compare third field sequences obtained by MinION, with Sanger sequence–based typing showing a 100% concordance between both data sets. In addition, the first validation panel was used to set the acceptance criteria for the use of MinION in a routine setting. The acceptance criteria were subsequently confirmed with the second validation panel. In summary, the present study describes validation and implementation of nanopore sequencing HLA class I typing method and illustrates that nanopore sequencing technology has advanced to a point where it can be used in routine diagnostics with high accuracy. Human leukocyte antigen (HLA) is the human major histocompatibility complex, a group of genes comprising the most polymorphic loci in the human genome.1Leffler E.M. Gao Z. Pfeifer S. Ségurel L. Auton A. Venn O. Bowden R. Bontrop R. Wall J.D. Sella G. Donnelly P. McVean G. Przeworski M. Multiple instances of ancient balancing selection shared between humans and chimpanzees.Science. 2013; 339: 1578-1582Crossref PubMed Scopus (190) Google Scholar The HLA genes are encoded within the short arm of the human chromosome 6, and they are grouped by both function and morphology into two general classes, HLA classes I and II. The hyperpolymorphism of HLA is demonstrated in the Immuno-Polymorphism Database–ImMunoGeneTics (IPD-IMGT)/HLA database,2Robinson J. Halliwell J.A. Hayhurst J.D. Flicek P. Parham P. Marsh S.G.E. The IPD and IMGT/HLA database: allele variant databases.Nucleic Acids Res. 2015; 43: D423-D431Crossref PubMed Scopus (1449) Google Scholar which currently lists 18,691 class I and 7065 class II HLA alleles (release 3.38.0). The nucleotide polymorphism is reflected in the protein polymorphism, which allows the HLA class I or class II molecules to present a wide variety of intracellular and extracellular antigens, respectively. HLA polymorphism enables the immune system to respond to a large variety of pathogens and diseases, but generates challenges in performing solid organ and stem cell transplantation requiring HLA typing. Both solid organ and stem cell transplantation involve the introduction of nonself tissue to the body, and have the inherent risk of adverse immune response. In solid organ transplantation, mismatched donor HLA can induce the production of donor-specific anti-HLA antibodies, which can bind to the HLA on the transplanted tissue,3Patel R. Terasaki P.I. Significance of the positive crossmatch test in kidney transplantation.N Engl J Med. 1969; 280: 735-739Crossref PubMed Scopus (1218) Google Scholar triggering antibody-mediated organ rejection. Although matching of patient and donor HLA alleles may not be possible because of low organ availability, high-resolution typing of HLA alleles identifies the amino acid sequence and consequent epitope structure of the HLA molecule, providing insight in the targets of the anti-HLA antibodies.4Duquesnoy R.J. Human leukocyte antigen epitope antigenicity and immunogenicity.Curr Opin Organ Transplant. 2014; 19: 428-435Crossref PubMed Scopus (51) Google Scholar,5El-Awar N. Jucaud V. Nguyen A. HLA epitopes: the targets of monoclonal and alloantibodies defined.J Immunol Res. 2017; 2017: 3406230Crossref PubMed Scopus (31) Google Scholar In stem cell transplantation, T cells in the donor allograft may recognize the HLA-peptide complexes expressed on recipient tissue as nonself, triggering global immune activation and leading to graft-versus-host disease. High-resolution HLA typing for stem cell transplantation is especially critical, because allele mismatches are linked with sometimes fatal adverse effects.6Fürst D. Müller C. Vucinic V. Bunjes D. Herr W. Gramatzki M. Schwerdtfeger R. Arnold R. Einsele H. Wulf G. Pfreundschuh M. Glass B. Schrezenmeier H. Schwarz K. Mytilineos J. High-resolution HLA matching in hematopoietic stem cell transplantation: a retrospective collaborative analysis.Blood. 2013; 122: 3220-3229Crossref PubMed Scopus (145) Google Scholar,7Mayor N.P. Hayhurst J.D. Turner T.R. Szydlo R.M. Shaw B.E. Bultitude W.P. Sayno J.-R. Tavarozzi F. Latham K. Anthias C. Robinson J. Braund H. Danby R. Perry J. Wilson M.C. Bloor A.J. McQuaker I.G. MacKinnon S. Marks D.I. Pagliuca A. Potter M.N. Potter V.T. Russell N.H. Thomson K.J. Madrigal J.A. Marsh S.G.E. Recipients receiving better HLA-matched hematopoietic cell transplantation grafts, uncovered by a novel HLA typing method, have superior survival: a retrospective study.Biol Blood Marrow Transplant. 2019; 25: 443-450Abstract Full Text Full Text PDF PubMed Scopus (63) Google Scholar High-resolution HLA typing has further applications in the area of drug hypersensitivity,8Fan W.L. Shiao M.S. Hui R.C. Su S.C. Wang C.W. Chang Y.C. Chung W.H. HLA association with drug-induced adverse reactions.J Immunol Res. 2017; 2017: 3186328Crossref PubMed Scopus (76) Google Scholar like abacavir and carbamazepine. HLA-B∗57:01 has been correlated with hypersensitivity to abacavir,9Illing P.T. Purcell A.W. McCluskey J. The role of HLA genes in pharmacogenomics: unravelling HLA associated adverse drug reactions.Immunogenetics. 2017; 69: 617-630Crossref PubMed Scopus (50) Google Scholar a drug commonly used to treat HIV, whereas HLA-B∗15:02 has been correlated with hypersensitivity to carbamazepine,10Leckband S.G. Kelsoe J.R. Dunnenberger H.M. George Jr., A.L. Tran E. Berger R. Muller D.J. Whirl-Carrillo M. Caudle K.E. Pirmohamed M. Clinical Pharmacogenetics Implementation Consortium guidelines for HLA-B genotype and carbamazepine dosing.Clin Pharmacol Ther. 2013; 94: 324-328Crossref PubMed Scopus (176) Google Scholar a drug used in epilepsy treatment. Typing for the HLA-B alleles informs the caregiver's choice of treatment methods and prevents adverse drug interactions. Furthermore, the HLA region is correlated with the highest numbers of human diseases in the human genome,11Dendrou C.A. Petersen J. Rossjohn J. Fugger L. HLA variation and disease.Nat Rev Immunol. 2018; 18: 325-339Crossref PubMed Scopus (301) Google Scholar and high-resolution sequencing allows refinement of understanding of the function of HLA and its relationship to the causality of diseases. In recent decades, increased understanding of the HLA genetics, advances in DNA sequencing technology, and a general lack of allele-specific antisera have led to a shift in HLA typing method. Serologic methods have often been replaced or supplemented by DNA sequence–based methods, allowing analysis of HLA at the nucleotide level.12Erlich H. HLA DNA typing: past, present, and future.Tissue Antigens. 2012; 80: 1-11Crossref PubMed Scopus (159) Google Scholar Sanger sequencing was developed in 1977,13Sanger F. Nicklen S. Coulson A.R. DNA sequencing with chain-terminating inhibitors.Proc Natl Acad Sci U S A. 1977; 74: 5463-5467Crossref PubMed Scopus (52650) Google Scholar and Sanger sequence–based typing (SSBT) has been the gold standard for HLA allele assignment for many years. However, with the increase in available HLA allele sequences, the heterozygous Sanger sequencing approach became more and more cumbersome, because of the increase in ambiguous typing results, which needed additional sequencing to resolve. This problem could be circumvented by group-specific full-length amplification and separate sequencing of the alleles,14Voorter C.E.M. Palusci F. Tilanus M.G.J. Sequence-based typing of HLA: an improved group-specific full-length gene sequencing approach.in: Beksaç M. Bone Marrow and Stem Cell Transplantation. Springer New York, New York, NY2014: 101-114Crossref Scopus (50) Google Scholar albeit this requires a preceding low-resolution typing to determine the allele groups. Technological advances led to next-generation sequencing (NGS), intended to be faster and cheaper than Sanger-based sequencing and with the huge advantage of separate allele sequencing. These technologies, including reversible terminator (Illumina, San Diego, CA)15Bentley D.R. Balasubramanian S. Swerdlow H.P. Smith G.P. Milton J. Brown C.G. et al.Accurate whole human genome sequencing using reversible terminator chemistry.Nature. 2008; 456: 53-59Crossref PubMed Scopus (2529) Google Scholar and semiconductor (Ion Torrent Systems, Thermo Fisher Scientific, Waltham, MA)16Merriman B. Rothberg J.M. Progress in ion torrent semiconductor chip based sequencing.Electrophoresis. 2012; 33: 3397-3417Crossref PubMed Scopus (240) Google Scholar methods, are commonly used for HLA typing,17Monos D. Maiers M.J. Progressing towards the complete and thorough characterization of the HLA genes by NGS (or single-molecule DNA sequencing): consequences, opportunities and challenges.Hum Immunol. 2015; 76: 883-886Crossref PubMed Scopus (13) Google Scholar but have the disadvantage of short sequences, impairing with correct alignment and phasing of the alleles. Third-generation sequencing is the term used to describe a new era of sequencing technologies that are focused on the analysis of single molecules (ie, long stretches of DNA without the need to fragment the DNA into smaller pieces as is the case for NGS-based techniques). Third-generation sequencing includes single-polymerase molecule (Pacific Biosystems, Menlo Park, CA)18Eid J. Fehr A. Gray J. Luong K. Lyle J. Otto G. et al.Real-time DNA sequencing from single polymerase molecules.Science. 2009; 323: 133-138Crossref PubMed Scopus (2448) Google Scholar and nanopore-based sequencing (Oxford Nanopore, Oxford, UK)19Branton D. Deamer D.W. Marziali A. Bayley H. Benner S.A. Butler T. Di Ventra M. Garaj S. Hibbs A. Huang X. Jovanovich S.B. Krstic P.S. Lindsay S. Ling X.S. Mastrangelo C.H. Meller A. Oliver J.S. Pershin Y.V. Ramsey J.M. Riehn R. Soni G.V. Tabard-Cossa V. Wanunu M. Wiggin M. Schloss J.A. The potential and challenges of nanopore sequencing.Nat Biotechnol. 2008; 26: 1146-1153Crossref PubMed Scopus (1938) Google Scholar technologies. The MinION20Jain M. Olsen H.E. Paten B. Akeson M. The Oxford nanopore MinION: delivery of nanopore sequencing to the genomics community.Genome Biol. 2016; 17: 239Crossref PubMed Scopus (37) Google Scholar is a portable nanopore sequencing platform that generates ultralong reads and requires little initial equipment investment. MinION uses a flow cell that contains a membrane with a grid of embedded nanopores, each of which is capable of binding to a DNA molecule. An electrical potential difference between both sides of the membrane is applied, generating a current across the membrane. Single-stranded DNA is passing through the pore with help from an accompanying motor protein.21Lieberman K.R. Cherf G.M. Doody M.J. Olasagasti F. Kolodji Y. Akeson M. Processive replication of single DNA molecules in a nanopore catalyzed by phi29 DNA polymerase.J Am Chem Soc. 2010; 132: 17961-17972Crossref PubMed Scopus (171) Google Scholar As nucleotides pass through the pore, disruptions in the current and resulting electrical signal are measured by the MinION integrated circuits. This signal is characteristic of the bases that are present within the pore, because of the varying size and morphology of the nucleotides. Software tools can translate the electrical signal back to the original nucleotide sequence, in a process known as basecalling.22Wick R.R. Judd L.M. Holt K.E. Performance of neural network basecalling tools for Oxford Nanopore sequencing.Genome Biol. 2019; 20: 129Crossref PubMed Scopus (397) Google Scholar The MinION is capable of natively sequencing a piece of single-stranded DNA, in a process known as 1D sequencing. A quality improvement is provided by MinION 1D2 protocols, where the sample preparation results in double-stranded DNA with adapter proteins attached to both ends, allowing both strands of the DNA to be individually sequenced. The MinION basecallers pair two complementary single-stranded reads in silico, resulting in a single, higher-accuracy read. The quality of the basecalling of a MinION read is represented by the per-base Phred quality scores,23Ewing B. Green P. Base-calling of automated sequencer traces using phred, II: error probabilities.Genome Res. 1998; 8: 186-194Crossref PubMed Scopus (4887) Google Scholar which were averaged over the length of each 1D2 read to calculate the mean. The mean reported Phred score over all the reads was found to be 18.5, which corresponds to 98.6% read accuracy. For the hyperpolymorphic HLA genes, in which each nucleotide difference can actually account for another allele, a high level of accuracy is essential to obtain reliable results. With reaching this high level of accuracy, typing of the hyperpolymorphic HLA genes by this MinION approach came into the picture. The potential of using the MinION as sequencing platform for the analysis of HLA has been described previously,24Montgomery M.C. Liu C. Petraroia R. Weimer E.T. Using nanopore whole-transcriptome sequencing for human leukocyte antigen genotyping and correlating donor human leukocyte antigen expression with flow cytometric crossmatch results.J Mol Diagn. 2020; 22: 101-110Abstract Full Text Full Text PDF PubMed Scopus (12) Google Scholar, 25Matern B.M. Olieslagers T.I. Voorter C.E.M. Groeneweg M. Tilanus M.G.J. Insights into the polymorphism in HLA-DRA and its evolutionary relationship with HLA haplotypes.HLA. 2020; 95: 117-127Crossref PubMed Scopus (16) Google Scholar, 26Duke J.L. Mosbruger T.L. Ferriola D. Chitnis N. Hu T. Tairis N. Margolis D.J. Monos D.S. Resolving MiSeq-generated ambiguities in HLA-DPB1 typing by using the Oxford nanopore technology.J Mol Diagn. 2019; 21: 852-861Abstract Full Text Full Text PDF PubMed Scopus (11) Google Scholar, 27Lang K. Surendranath V. Quenzel P. Schofl G. Schmidt A.H. Lange V. Full-length HLA class I genotyping with the MinION nanopore sequencer.Methods Mol Biol. 2018; 1802: 155-162Crossref PubMed Scopus (15) Google Scholar, 28Liu C. Xiao F. Hoisington-Lopez J. Lang K. Quenzel P. Duffy B. Mitra R.D. Accurate typing of human leukocyte antigen class I genes by Oxford nanopore sequencing.J Mol Diagn. 2018; 20: 428-435Abstract Full Text Full Text PDF PubMed Scopus (20) Google Scholar, 29Chua E.W. Ng P.Y. MinION A novel tool for predicting drug hypersensitivity?.Front Pharmacol. 2016; 7: 156Crossref PubMed Scopus (6) Google Scholar, 30Ammar R. Paton T.A. Torti D. Shlien A. Bader G.D. Long read nanopore sequencing for detection of HLA and CYP2D6 variants and haplotypes.F1000Res. 2015; 4: 17Crossref PubMed Google Scholar, 31Tilanus M.G.J. The power of Oxford nanopore MinION in human leukocyte antigen immunogenetics.Ann Blood. 2017; 2: 12Crossref Google Scholar but up until now it has not been used in routine diagnostics. The current article describes the complete validation process of full-length HLA class I single-molecule sequencing and typing method using the Oxford Nanopore MinION and the implementation in the routine diagnostic setting. To validate the MinION approach for reliable HLA typing, quality standards and validation metrics needed to be generated. A panel of samples with known HLA high-resolution typing was sequenced and typed with the standard 1D2 MinION protocol. The initial panel consisted of 33 samples, which cover 70 of the 71 known class I allele groups, excluding HLA-B∗83, which was not available in our laboratory (Table 1). This panel was subjected to the MinION 1D2 approach, with a focus on comparing results with our Sanger sequencing results (Supplemental Table S1) and determining read quality and coverage statistics. For each sample in the initial validation, read coverage was measured on each MinION run, for each demultiplexed MinION barcode, for each HLA locus within a barcode, and for both HLA alleles at a locus.Table 1Samples Included in the Initial ValidationSample identifierHLA-A∗HLA-B∗HLA-C∗101:01:01:0130:01:0115:10:0142:01:0103:04:0217:01:01202:06:01:0102:06:01:0451:01:0159:01:0101:02:0114:02:01302:01:01:0152:01:0173:0107:01:0115:05:01401:0266:01:01:0158:01:0158:02:0103:02:02:0106:02:01502:01:0136:0115:03:01:0251:01:0101:02:0112:03:01602:01:0131:01:02:0115:01:01:0167:01:0107:02:01726:01:01:0130:02:01:0118:01:0140:01:0203:04:01:0105:01:01:01824:02:01:0132:01:01:0114:01:01:0118:01:0107:01:0108:02:01:02901:01:01:0131:01:02:0108:01:0140:01:0203:04:01:0107:01:011001:01:01:0103:01:01:0108:01:0145:01:0106:02:01:0307:01:011102:06:01:0130:02:01:0118:01:0139:0805:01:0107:02:011229:02:01:0169:01:01:0139:06:02:0155:01:0101:02:0107:02:011303:01:01:0566:01:01:0115:03:01:0252:01:01:0102:10:01:0112:02:02:011443:0174:01:0115:03:01:0244:0302:10:01:0108:04:011502:01:0134:01:0140:02:0156:02:0101:02:0115:02:01:011630:01:0133:01:01:0153:01:0181:0104:01:0108:04:011702:01:0124:02:01:0144:02:01:0149:0305:01:01:0207:01:011803:01:01:0525:01:01:0137:01:01:0147:01:01:0306:02:01:011902:03:0102:07:0138:02:0146:01:0101:02:0107:02:012001:01:01:1102:01:0135:04:0182:01:01:0103:02:02:0104:01:012102:01:0144:0950:01:01:0105:01:01:022224:02:01:0126:02:0140:06:01:0154:01:0101:02:0108:01:01:012324:02:0133:03:0115:07:01:0215:16:01:0203:03:0114:02:01:022426:01:01:0174:01:0181:0178:01:01:0216:01:0118:012523:01:01:0132:01:01:0141:02:0144:03:0104:01:0117:03:012611:01:01:0124:02:0127:0648:01:0101:02:0108:01:01:012701:01:38L02:01:0115:17:01:0157:01:01:0106:02:0107:01:022802:01:0125:01:01:0115:78:0138:01:01:0103:04:01:0112:03:01:012930:0980:01:01:0207:02:0181:01:0107:02:0118:023011:01:01:0168:01:02:0140:01:0255:01:0101:02:0107:02:013124:02:01:0129:01:01:0107:05:0127:02:01:0402:02:0215:05:023224:1733:03:0107:02:0115:02:01:0107:02:01:0308:01:01:013302:01:01:0124:02:0107:02:0113:02:01:0106:02:01:0107:02:01Samples were selected to cover 70 of the 71 known HLA-A, HLA-B, and HLA-C allele groups. HLA-B∗83 was not included, as it was not available in our laboratory. Open table in a new tab Samples were selected to cover 70 of the 71 known HLA-A, HLA-B, and HLA-C allele groups. HLA-B∗83 was not included, as it was not available in our laboratory. Acceptance criteria were defined in the initial validation panel, and the defined criteria were verified by sequencing and HLA typing a second panel of 67 samples (402 alleles) from our laboratory in a secondary validation phase parallel to Sanger sequencing (Supplemental Table S2). The samples were sequenced and analyzed using the combined analysis approach, with optimizations based on the initial validation. The samples were also typed using full-length allele-specific SSBT approach.14Voorter C.E.M. Palusci F. Tilanus M.G.J. Sequence-based typing of HLA: an improved group-specific full-length gene sequencing approach.in: Beksaç M. Bone Marrow and Stem Cell Transplantation. Springer New York, New York, NY2014: 101-114Crossref Scopus (50) Google Scholar Typing results from MinION were compared with the SSBT results. Accuracy of HLA allele assignment of these samples was considered on the basis of criteria defined in the initial validation. The MinION sequencing and analysis method is outlined in Figure 1. The procedure starts with the PCR amplification of 300 ng DNA, purified and isolated from peripheral blood samples and according to the descriptions outlined in Voorter et al.14Voorter C.E.M. Palusci F. Tilanus M.G.J. Sequence-based typing of HLA: an improved group-specific full-length gene sequencing approach.in: Beksaç M. Bone Marrow and Stem Cell Transplantation. Springer New York, New York, NY2014: 101-114Crossref Scopus (50) Google Scholar At a later time DNA isolated from buccal swabs was tested for MinION sequencing with comparable results. The full-length gene-specific amplification primers are located in the 5′ and 3′ untranslated regions of the HLA-A, HLA-B and HLA-C genes (Table 2). To sequence several samples simultaneously, specific Oxford Nanopore tag and barcode sequences were added to the primers. The tag sequence coordinates the basecalling and demultiplexing software, and the barcode sequence is used to sort multiplexed samples. After amplification, presence of PCR products was confirmed by agarose gel electrophoresis, and PCR products were purified using CleanPCR magnetic beads (GC Biotech, Waddinxveen, the Netherlands). In addition, by using a bead versus DNA ratio of 1:1 during the purification, primer dimers were removed simultaneously. Subsequently, up to nine samples (27 PCR products) were equimolar pooled, with equal distribution between loci, to a total quantity of 1300 ng DNA.Table 2Amplification PrimersGeneDirectionSequenceIMGT/HLA gDNA positionHLA-AForward5′-GGATACTCACGACGCGGAC-3′−137 to −119HLA-AReverse5′-GGGAGCACAGGTCAGCGTGGGAAG-3′3075 to 3098HLA-BForward5′-GGCAGACAGTGTGACAAAGAGGC-3′−420 to −398HLA-BReverse5′-CTGGGGAGGAAACACAGGTCAGCATGGGAAC-3′3040 to 3070HLA-CForward5′-TCAGGCACACAGTGTGACAAAGAT-3′−327 to −304HLA-CReverse5′-TCGGGGAGGGAACACAGGTCAGTGTGGGGAC-3′3067 to 3098This table contains the gene-specific amplification primers. Only the sequence that complements the HLA untranslated region sequence is shown; the amplification primers also include a section of sequence containing MinION tag sequences, as well as DNA barcodes, as provided by Oxford Nanopore.gDNA, genomic DNA; HLA, human leukocyte antigen; IMGT, ImMunoGeneTics. Open table in a new tab This table contains the gene-specific amplification primers. Only the sequence that complements the HLA untranslated region sequence is shown; the amplification primers also include a section of sequence containing MinION tag sequences, as well as DNA barcodes, as provided by Oxford Nanopore. gDNA, genomic DNA; HLA, human leukocyte antigen; IMGT, ImMunoGeneTics. Pooled samples were further prepared for MinION sequencing using the 1D2 sequencing kit (SQK-LSK308; Oxford Nanopore) and following manufacturer's protocol with a few minor adjustments. In short, the amplicon strands are end repaired and dA tailed using the NEBNext End Repair/dA tailing module (New England Biolabs, Ipswich, MA). These end-repaired amplicons were purified using CleanPCR beads and ligated with 1D2 adapters, which allows the nanopore to capture the complement strand immediately after the template. After another purification step, sequencing adapters were ligated onto the amplicons, which ensures that the DNA strands can enter the nanopore. The MinION, with an attached flow cell, was connected to a computer and quality control was performed, which checks for available and active nanopores, the R9.5 flow cell was primed, and 75 μL of the prepared 1D2 library was loaded into the flow cell for sequencing. The sequencing run was performed for 16 hours and was controlled by MinKNOW software version 18.07.2 (Oxford Nanopore), which collects the 1D read data. Basecalling was performed initially using Albacore software version 2.3.1 and later updated to Guppy software version 3.2.4 (both from Oxford Nanopore). The basecaller first converts electrical signal from the 1D reads into a nucleotide sequence, and the sequences within 1D reads from complementary strands are subsequently paired and combined into higher-accuracy 1D2 reads. 1D2 reads containing the MinION tag and barcode sequences were demultiplexed initially by Porechop version 0.2.3 (https://github.com/rrwick/Porechop, last accessed December 1, 2019), later updated to the same Guppy software. Porechop/Guppy removes the non-HLA tag and barcode sequences from the reads and sorts the reads into FASTQ files corresponding to each barcode sequence. Data analysis and interpretation were performed by a combined approach, using two separate software packages: JSI SeqPilot SeqNext (NGS) HLA module version 4.4.0 (JSI, Ettenheim, Germany) and GenDx NGSengine version 2.11.0.11444 (GenDx, Utrecht, the Netherlands). Both are configured to ignore the regions containing amplification primers. Analysis of the sorted HLA 1D2 read data was performed in both software packages independently, and any discrepancies between the two programs were analyzed in detail. The estimated duration of time for the MinION sequencing procedure is as follows for the different parts: PCR amplification, 5 hours; cleanup, pooling, and library preparation, 5 hours; and MinION 1D2 sequencing, 16 hours. Because basecalling and analysis time is highly dependent on the power of the computer used and the experience of the user, it is not possible to make any time estimation for this part. HLA sequences obtained by MinION sequencing were compared with typing results from full-length group-specific SSBT,14Voorter C.E.M. Palusci F. Tilanus M.G.J. Sequence-based typing of HLA: an improved group-specific full-length gene sequencing approach.in: Beksaç M. Bone Marrow and Stem Cell Transplantation. Springer New York, New York, NY2014: 101-114Crossref Scopus (50) Google Scholar which is considered the gold standard for HLA typing in our laboratory. In brief, DNA was isolated and amplified using group- and allele-specific primers for the HLA class I genes. The DNA product was sequenced on a Sanger 3730 analyzer (Applied Biosystems, Foster City, CA), and sequence analysis and allele calling were performed using the JSI SeqPilot SeqHLA module. For the initial validation panel, consisting of 33 samples covering 70 of the 71 known HLA class I allele groups (Table 1), HLA-A, HLA-B, and HLA-C were successfully amplified and sequenced using the MinION sequencing method. Data were analyzed and interpreted using two separate software packages: JSI SeqPilot SeqNext (NGS) HLA module and GenDx NGSengine. Because HLA analysis software programs specifically designed for MinION data were not yet available, a combination of two different HLA analysis programs was chosen, both able to deal with data from all common NGS platforms and kits. Comparisons of the typing results with the results of Sanger full-length class I HLA sequence–based typing (SSBT) reveals that the MinION 1D2 sequencing protocol and redundant analysis approach was 100% concordant with the SSBT typings to third-field resolution (Supplemental Table S1). Although most of the samples were correctly typed to three fields in both software tools immediately, some manual interpretation was necessary in a small percentage of the typing results. Because of the presence of homopolymer stretches, in 14.1% of the cases, one of the nucleotides within the homopolymer sequence was ignored in the NGSengine software, resulting in an ambiguous typing result (Supplemental Table S1). This nucleotide was, however, correctly identified with the other analysis program, resulting in a correct allele assignment and therefore a correct final HLA typing. For three alleles (1.5%), a discrepancy was observed between the allele call obtained with NGSengine and SeqPilot NGS. In these cases, SeqPilot correctly assigned the allele to two fields, but a region in the introns containing a short tandem repeat (STR) sequence was misaligned, resulting in misalignment of the exon and therefore incorrect third-field allele call. These alleles were, however, correctly typed to the third field by NGSengine, and manual inspection of the analysis details easily resolved the discrepancy. In a single sample, NGSengine was unable to assign an allele for HLA-A (Supplemental Table S1), whereas the same MinION sequence data in SeqPilot NGS gave the correct typing without any problems. Repeating the sample did not solve the problem, whereas other samples with the same typing did not demonstrate this problem. Another problem observed during this initial validation was co-amplification of HLA-Y with HLA-A in samples that were positive for HLA-A∗30, HLA-A∗31, HLA-A∗33, and HLA-A∗