Search

Aim HLA-E has been reported to impact the outcome of hematopoietic stem cell transplantations (HSCT). Starting July 2017, DKMS added HLA-E to the default typing profile for newly enrolled potential donors for HSCT. Since then, DKMS Germany has typed more than 390,000 samples for HLA-E together with the HLA loci A, B, C, DRB1, DQB1, and DPB1. In this work, we present 7-locus haplotype and allele frequencies for donors from Germany (N = 325,955), Turkey (N = 10,649), Poland (N = 4144), Russia (N = 2896) and Italy (N = 2035). Methods HLA-E is typed by our well-establish high-throughput approach relying on an amplicon spanning exons 2 and 3. Amplicons are sequenced by Illumina MiSeq or HiSeq 2500 instruments without fragmentation, preserving phase information. Reads are analysed by our in–house typing software neXtype. Typing results comprise the high-resolved alleles HLA-E*01:01:02, 01:03:03, 01:04, 01:05, 01:07 and 01:08 N and three G-groups: HLA-E*01:01:01G, 01:03:01G and 01:03:02G. Upon enrollment with DKMS Germany, potential stem cell donors provide information on their self-assessed ethnic background by statement of their respective country of origin. Using our open-source software Hapl-o-Mat, we computed frequencies on g-group resolution level for both, alleles and 7-loci haplotypes. Results The most common allele group in all five populations is HLA-E*01:01 g with a fairly similar frequency (f Germany = 55.5%, f Turkey = 53.4%, f Poland = 56.9%, f Russia = 57.0%, f Italy = 53.0%). Frequencies for HLA-E*01:03 g are (f Germany = 44.4%, f Turkey = 46.6%, f Poland = 43.1%, f Russia = 43.0%, f Italy = 46.9%). In all populations, except Russia, HLA-E*01:05 is the third most common allele. In the Russian population only the two dominant alleles were observed. Conclusions Our approach provided reliable HLA-E typing data for a large number of samples. In all considered populations, identical allele groups are dominant with similar frequencies.

Aim HLA typing by Next Generation Sequencing (NGS) enables three to four field typing of the HLA-A,-B,-C,-DRB1345,-DQA1,-DQB1,-DPA1 and -DPB1 loci in one step, making it easier to identify potential haematopoietic cell family donors who are HLA matched. This study presents three cases where HLA-DP typing, as part of the standard NGS protocol, distinguished HLA matched siblings from haploidentical or non–haplotype matched siblings. Methods HLA-A,-B,-C,-DRB1345,-DQA1,-DQB1,-DPA1 and -DPB1 typing by NGS was carried out using the Illumina TruSight HLA v2 Sequencing Panel on the Illumina MiSeq instrument, with Assign 2.1 TruSight HLA Analysis software (Illumina, San Diego, CA, USA). The study includes 58 adult patients and 124 siblings; and 10 paediatric patients, both parents and 19 siblings. Results NGS typing showed 24 (41%) of the adult patients had at least one HLA-A,-B,-C,-DRB1345,-DQA1,-DQB1,-DPA1 and -DPB1 matched sibling, while 43 (74%) had a presumed haploidentical sibling, (no parents were available to confirm haplotypes). In one case, the adult patient and all five siblings were fully matched at HLA-A,-B,-C,-DRB1345,-DQA1 and -DQB1. However, three siblings had different DPB1 alleles to the patient, indicating they were haploidentical, rather than being HLA matched. In another case, two siblings were HLA-A,-B,-C,-DRB1345,-DQA1, and -DQB1 identical to each other, but differed at DPB1. The DP difference meant only one sibling was haploidentical to the adult patient. With the paediatric cases, there was one family where a crossover appeared to have occurred between HLA-DQ and HLA-DP in one of the two healthy siblings, although neither was a haplotype match to the patient. Conclusions With the implementation of NGS typing, HLA-DPA1 and -DPB1 typing is now routine in our laboratory for all haematopoietic cell transplant workups. This has shown some siblings, who previously would have been thought to be HLA matched (based on HLA-A,-B,-C,-DRB1345,-DQA1 and -DQB1), were only haplotype matched, while other siblings thought to be haplotype matched were not, once HLA-DP typing was considered. Therefore it is beneficial to include HLA-DP typing for all haematopoietic cell transplant workups.

Aim HLA genotyping by next-generation sequencing has matured from a proof-of-principle concept to the development of several commercially available kits all within the space of a few years. We evaluated several kits against our own in-house developed method to determine the best-fit solution for our clinical lab. Methods Nine clinical samples were compared using kits from Omixon Holotype X2 24/7, Illumina HLA TruSight v2, Immucor MiaFora, and an in-house developed method. Directions were followed as indicated per commercial kit and all methods were performed manually. Sequencing was performed on the Illumina MiSeq and HLA assignments were made using each kit’s supplied software. The in-house method used Omixon’s Target software for analysis. Methods were evaluated for genotyping accuracy and ease of use. Results All methods gave comparable, unambiguous genotyping results (95.2–98.8%) but there were significant differences in the library construction process. Of the four methods, the Illumina HLA TruSight v2 required the fewest steps. In contrast to the other three kits, the Illumina method did not require the use of fluorometer (amplicon quantification), size-selection instrument (PippenPrep), or a real-time PCR instrument (final library quantification), which simplifies the library preparation process significantly. Both the MiaFora and the Illumina HLA TruSight use the 2 × 150 (300 cycle) MiSeq reagent kit and thus have shorter run times (20 h) in comparison to the Omixon and in-house methods that require use of the longer read length kits (2 × 250, 500 cycles) and thus have a longer run time (40 h). Conclusions All of the methods evaluated correctly genotyped a small set of samples although there were a few technical failures. The biggest difference between these methods is in the library preparation process. Depending upon the commercial method selected, additional budgeting and purchasing decisions may be required.

Aim We present the evaluation of two NGS-based HLA typing kits from Illumina, comparing TruSight kit version 1 with the newly released version 2 kit. Methods Following Illumina’s protocol, 26 clinical and two IHWG DNA samples were evaluated using both kits. Samples had previously been Sanger sequenced. Pooled libraries were run on the Illumina MiSeq. Version 1 required the use of the 2 × 250 (500 cycle) MiSeq Reagent Kit whereas, version 2 required the use of the 2 × 150 (300 cycle) kit. Fastq files were analyzed using Conexio’s Assign software, versions 1.0.0.729 and 2.0.0.889 respectively. Results The version 2 kit outperformed the version 1 kit in several respects: there were fewer failures, fewer two-field ambiguities for B and DRB1 genes, and fewer problems with coverage and quality. Improvements in typing resolution are due to changes in primer locations. In addition, changes to the library preparation procedure simplified the process. In version 2, sample amplicons are now pooled much earlier in the process resulting in less pipetting and fewer consumables required. The change from the 500 to 300 cycle MiSeq kit resulted in time savings. The 500 cycle kit required 40 h of instrument run time while the 300 cycle kit only required 20 h of run time. In version 1, 88.9% of all samples typed correctly and unambiguously (2-field) whereas 97.8% of samples did so in version 2. Conclusions The improvements and change to the 300 cycle reagent kit enabled us to take six samples from genomic DNA to high-resolution HLA genotyping results in 48 h. Illumina has made numerous improvements to their HLA TruSight HLA NGS kit which has made it a viable option for clinical implementation.

Aim Next-generation sequencing (NGS) technology generates sequences from single DNA molecules that enables unique identification of the paternal and maternal alleles. The NGS technology has great potential in the application of HLA typing for diagnostic purposes and generating reliable, unambiguous HLA typing results in a high-throughput fashion. Here we present data obtained for 11 HLA loci using a HLA typing assay dedicated for high-throughput NGS-based HLA typing on the Illumina MiSeq. The implementation of this high-throughput NGS strategy would be highly advantageous to obtain robust high-resolution HLA typing results for registry labs and routine diagnostic purposes. Method A large genomic DNA reference panel, including UCLA reference samples, were selected for high-throughput NGS HLA typing. The whole gene of HLA-A, -B, -C, and the essential genomic regions of HLA-DRB1, -DRB345, -DQA1, -DQB1, -DPA1 and -DPB1 were PCR-amplified in a single PCR for each locus. Amplicons of each sample were pooled and processed using our NGSgo® DNA library preparation workflow for Illumina MiSeq. During library preparation the samples were fragmented, barcoded and multiplexed. Finally paired-end sequencing (2x150 cycles) was performed on the Illumina MiSeq platform using a standard flow cell and data was analysed using our HLA typing software NGSengine®. Results Paired-end NGS data was obtained for 11 HLA loci (class I and II) of most genomic DNA reference samples in a single Illumina MiSeq sequencing run. The NGS data covered the complete HLA region of interest by highly accurate measures for all HLA loci tested. All HLA alleles were effectively separated and, in most cases, could be phased throughout the gene. For the majority of samples, unambiguous third- and fourth-field resolution typing were obtained for all loci that are in concordance with the second-field or higher reference types. Conclusions The high-throughput NGSgo® strategy that we have developed demonstrates to be a powerful method to perform high-resolution HLA typing by means of NGS in a robust and reliable manner. L. Van de Pasch: Employee; Company/Organization; GenDx. M. Bacelar: Employee; Company/Organization; GenDx. C. Rebel: Employee; Company/Organization; GenDx. R. Van Aard: Employee; Company/Organization; GenDx. R. Kooter: Employee; Company/Organization; GenDx. E. Rozemuller: Stock Shareholder; Company/Organization; GenDx. M. Penning: Employee; Company/Organization; GenDx. E. Bouwmans: Employee; Company/Organization; GenDx. N. Westerink: Employee; Company/Organization; GenDx.

Aim Given the high throughput, better coverage, and free of phase ambiguity with HLA typing using next generation sequencing (NGS) technology, we decided to test one of the commercially available kits (Holotype HLA X4 ) offered by Omixon. Methods Total of 32 samples (27 PBMC derived patient DNA samples and 5 UCLA survey DNA samples) were sequenced by NGS for HLA-A, B, C, DRB1 and DQB1 using Holotype HLA X4. All samples were previously typed at high-resolution by combinations of SBT/SSP. This includes 23 HLA-A, 28 B, 21 C, 28 DRB1, 14 DQB1 CWD2.0 alleles and also 2 non-CWD and 2 novel alleles. With Omixon HLA X4 kit, both individual PCR amplicon barcode and pooled amplicon barcode strategies were evaluated on Illumina MiSeq analyzer. Results Long-range PCR gave robust amplifications on all samples for HLA-A, B, C (full coverage) and DRB1 (exon 2–4). DQB1 set 2 PCR (exon 2–5) had a failure rate of 3% (1/32) and DQB1 set 1 PCR (exon 1–4) had a failure rate of 21.9% (7/32, 5 of them from UCLA survey samples). The accuracy of NGS typing was 100% concordant with previous typing for HLA-A, B, C in both PCR amplicon individually barcoded and pooled amplicon barcoded format. Accuracy dropped to 97% for DRB1 and DQB1 assignment in both formats due to technical errors (1/32) or failed DQB1 PCR (1/32). Non-CWD alleles as well as novel alleles were correctly identified. No ambiguities were present in any of the final typing assignments. Using default quality control matrix of 15 parameters from the HLA Twin software, 4/158 resolved HLA alleles gave “red” light in the two traffic light system for amplicon individually barcoded samples, compared to 7/158 alleles in pooled amplicon barcoded samples. Regardless, none of the sample with “red” flagged QC parameter compromised the final HLA typing assignment. Conclusions Our experience with Omixon Holotype HLA X4 kit indicates that HLA-NGS typing indeed has advantages over Sanger SBT method. The Omixon NGS library construction is easy to perform without noticeable bias introduction. The beta version HLA Twin analysis software is user friendly & performs well with both amplicon individually barcoded as well as pooled amplicon barcoded samples with extensive data quality monitoring. However, the robustness of DQB1 PCR requires improvement & the dual algorithm of HLA Twin software needs more transparency.

Aim Although the Luminex system for HLA typing provides a convenient, easy to perform laboratory workflow, limitations restricting the ability to routinely obtain both breadth (number of loci) and depth (allele resolution) of data are inherent in the system. If high-resolution data is required, high volume laboratories can minimize costs effectively but for clinical laboratories in direct support of hematopoetic cell transplants (HCTs) costs of dye-terminator sequence-based typing are relatively high. In order to improve the economy of obtaining both broad and high-resolution data for our laboratory, we tested a next generation sequencing (NGS) method for HLA typing. Methods The method uses the Illumina MiSeq device combined with commercial software providing a convenient workflow for data generation and analysis. Data for HLA-A, B, C, exons 1-7 are obtained with phase determined by a combination of overlapping sequences and database lookup. Data for DRB1/3/4/5, DQA1, DQB1, DPA1, and DPB1 are obtained from exons 1-4 with phase between exons determined by coincidence with established databases. Results This approach was tested with parallel typing using the NGS and Luminex systems for over 200 samples to date. We intend to continue with parallel typing during the next few months for a total of about 500 samples for our final report. Overall results demonstrated unambiguous 3-field (six-digit) typing for all HLA class I and II with less than 1% failure rate and complete coincidence with Luminex data. Improvements in both accuracy and resolution are provided by the NGS system, and laboratory effort required is similar. Software and data storage are performed on the cloud, providing rapid analysis, ease of use, and freedom from on site systems support. Conclusion We detail the comparative typing results and critically examine the NGS workflow towards establishing the efficacy and economy of this NGS approach in laboratories supporting HCTs.

Aim Currently applied technologies for HLA sequencing-based typing (SBT) often show ambiguous results, most of which are caused by cis–trans allele combinations. In contrast, next-generation sequencing (NGS) allows for reads that originate from a single molecule, meaning that the paternal and maternal alleles can uniquely be identified. As NGS can phase single molecules throughout the gene, HLA typing by means of NGS will yield typing results without ambiguities. The implementation of NGS technology for routine diagnostic purposes would therefore be highly advantageous to obtain reliable, unambiguous HLA typing results. Methods We developed a complete NGS workflow for high-throughput HLA typing compatible with the Illumina NGS platforms. The NGS workflow, including full-length HLA locus amplification, library preparation, and sequencing analysis, is fully optimized to facilitate high-resolution identification of HLA alleles. We generated HLA locus specific amplicons for the class I loci (HLA-A, -B and -C) and class II loci (HLA-DRB1, -DRB3, -DRB4, -DRB5, -DPA1, -DPB1, -DQA1, and -DQB1) on 96 genomic DNA reference samples, using our optimized NGS workflow compatible with the Illumina MiSeq platform. Samples and loci were pooled and run on the MiSeq. Sequence data was analyzed using the NGS HLA typing software NGSengine®. Results Paired-end NGS data was obtained for 11 class I and class II HLA loci pooled for 96 gDNA reference samples in one MiSeq sequencing run. The NGS data covered the complete HLA region of interest by highly accurate measures for all HLA loci tested. All HLA alleles were effectively separated and, in most cases, could be phased throughout the gene, resulting in unambiguous high-resolution HLA typings. High-quality data was obtained for all loci in agreement with the reference SBT allele assignments as provided by Sanger SBT. Conclusion The NGS workflow we have developed demonstrates that it is feasible to perform HLA typing by means of NGS in a reliable, accurate and high-throughput manner that in future will be suitable for routine diagnostic purposes. J. Adema: Employee; Company/Organization; GenDx. M. Bacelar: Employee; Company/Organization; GenDx. R. Kooter: Employee; Company/Organization; GenDx. F. Ruzius: Employee; Company/Organization; GenDx. E.H. Rozemuller: Employee; Company/Organization; GenDx. Stock Shareholder; Company/Organization; GenDx. M.T. Penning: Employee; Company/Organization; GenDx. L. van de Pasch: Employee; Company/Organization; GenDx. N. Westerink: Employee; Company/Organization; GenDx.

Aim NGS is well-suited for HLA typing as it can deliver highly accurate and unambiguous results. A simplified protocol has been developed for use with the Illumina MiSeq and has been subjected to an inter- Methods: Six laboratories, each with varying levels of NGS experience, participated in this double-blinded study. The same 16 samples were typed at HLA-A, B, C, DRB1, and DQB1 by each lab. The protocol consisted of LR-PCR, library prep, and paired-end 250 bp sequencing. Two indexing strategies were employed: (1) Locus-specific indexing by which each locus was tagged uniquely and (2) sample-specific indexing whereby all 5 loci for a sample were pooled prior to library prep. Sequence analysis was performed with Target HLA (Omixon) and NGSengine (GenDx) to assess complementarity of two analysis algorithms. Results Six sequencing runs yielded an average output of 7.8 Gb per run. The average number of sequence reads per library was 387,813. However, analysis was limited to 40K reads for the locus-indexed libraries and 200K reads for the sample-indexed libraries (amplicon pools) resulting in an average coverage of 1444. A sufficient number of reads for genotype analysis were obtained for 98.4% of libraries. Genotype accuracy is shown in Table 1. Reproducibility between labs was 99.7%. The only cause of discordance between labs was cross-contamination of a single amplicon. Conclusion The protocol developed for the MiSeq is reproducible, highly accurate, and its simplicity makes it amenable to clinical testing. The ability of the software to decipher reads from multiple loci in a single library demonstrates the potential for using an amplicon pooling approach that further simplifies sample prep and reduces expense. As demonstrated in this study the protocol is suitable for moderate test volume. However, since 100 samples can be typed in a single run using this approach. A. Berces: Employee; Company/Organization; Omixon, Inc. T. Hague: Employee; Company/Organization; Omixon, Inc. G. Horvath: Consultant; Company/Organization; Omixon, Inc. R. Kooter: Employee; Company/Organization; GenDx . W. Mulder: Stock Shareholder; Company/Organization; GenDx . M. Penning: Employee; Company/Organization; GenDx. K. Rigo: Employee; Company/Organization; Omixon, Inc. E. Rozemuller: Stock Shareholder; Company/Organization; GenDx.

Aim Next-generation sequencing (NGS) technology has great potential for the future application of HLA typing in routine diagnostic purposes. NGS sequencing has the advantage that it allows the sequencing of single molecule DNA sequences, meaning that the paternal and maternal alleles can be uniquely identified. As NGS can phase single molecules throughout the gene, HLA typing by means of NGS will yield typing results without ambiguities. Currently, various NGS platforms are available on the market; each platform is based on different technology and each technology has their own specific requirements in order to make them compatible for HLA typing. It is our aim to develop multiple NGS workflows that are each compatible to a specific NGS platform, such as the Illumina MiSeq, the IonTorrent PGM, and the PacBio RS Sequencer. Method Here we present our latest developments on the IonTorrent PGM platform specific workflow, which is designed for the high-resolution typing of HLA loci. The new IonTorrent workflow for HLA typing includes several consecutive steps, from the most crucial step of whole HLA locus amplification to fragmentation, adaptor ligation, suppression PCR, clonal amplification by emulsion PCR, enrichment of positive ion sphere particles and sequencing on the IonTorrent PGM platform. Results We present the application of the IonTorrent workflow for 24 samples from standardized genomic DNA reference panels. GenDx adapters are developed and applied, enabling indexing of multiple samples. The NGS data is analyzed using the NGS platform-independent NGSengine® software, developed for NGS HLA sequence analysis (GenDx). Conclusion The availability of different NGS workflows allows the end-user to choose most preferred SBT-NGS HLA typing method that fulfills their laboratory application needs and requirements. This GenDx NGS workflow for the IonTorrent PGM presented here, is a powerful method to perform HLA typing by means of NGS in a reliable, accurate and high-throughput manner. L. Krol: Employee; Company/Organization; GenDx. J. Adema: Employee; Company/Organization; GenDx. R. Kooter: Employee; Company/Organization; GenDx. F. Ruzius: Employee; Company/Organization; GenDx. E.H. Rozemuller: Employee; Company/Organization; GenDx. Stock Shareholder; Company/Organization; GenDx. L. van de Pasch: Employee; Company/Organization; GenDx. N. Westerink: Employee; Company/Organization; GenDx.

Aim Next Generation Sequencing (NGS) technologies are more frequently used to determine HLA genotypes. Since all sequence reads originate from a single molecule, in most cases the nucleotide sequences of the paternal and maternal alleles are determined separately. During PCR amplification the DNA template is amplified, resulting in copies of the original template. However, this process is not perfect, leading to wrong bases in individual molecules. These errors can be identified in NGS, where it is considered noise. In addition, hybrid molecules are generated which represents reactions which start at one molecule, but at some point the product re-anneals to another template, and continues to elongate. Especially if a template originates from a heterozygous sample, and covers multiple positions where the maternal and paternal allele differ, these hybrid molecules can be detected by NGS. In this study we quantified the hybrid molecules as hybrid reads by NGS, and applied it as a quality value for the PCR amplification. Methods HLA gene amplicons were generated with GenDx’ NGSgo-AmpX in independent experiments in different labs. NGS sequencing was done on Illumina MiSeq, Ion Torrent PGM and PacBio RS. NGSengine, a software package to perform HLA typing based on NGS data was used to quantify the number of hybrid reads in several different HLA gene amplification systems. The amount of hybrid reads was compared to other HLA gene amplification strategies. Results Detailed analyses is currently be carried out. We will present the effect of cycling conditions on the generation of hybrid molecules. Furthermore we will show to what extend this phenomenon is related to one or more platforms. We will come up with measures such that this may be exploited as a quality parameter for NGS HLA Typing. Conclusions NGSengine allows for easy investigation of hybrid reads. Hybrid reads can be used as quality control for amplification robustness. R. Kooter: Employee; Company/Organization; GenDx. M.T. Penning: Employee; Company/Organization; GenDx. W. Mulder: Employee; Company/Organization; GenDx. Stock Shareholder; Company/Organization; GenDx. E.H. Rozemuller: Employee; Company/Organization; GenDx. Stock Shareholder; Company/Organization; GenDx.

Aim We have previously considered transposon-and sonication-based library preparation methods for next generation sequencing of HLA-A, B, C, DRB1 and DQB on the Illumina MiSeq, and found that they are not suitable for routine clinical work. Here, we evaluated reagents from NEB, NEBNext dsDNA Fragmentase and NEBNext DNA Library Prep Master Mix, to generate sequencing libraries for clinical testing. Methods We performed long-range PCR for HLA-A, B, C, DRB1 and DQB1 on two IHWG samples using a combination of in-house and commercial primers. The amplified products ranged from 3900 to 4800 bp in length. The class I primers were designed to cover all exons, introns and UTRs whereas the class II primers covered exon 2 to the 3′UTR. Samples were enzymatically fragmented using NEBNext dsDNA Fragmentase which is composed of two different endonucleases isolated from V. vulnificus and T7. Length of incubation time of the amplicon and Fragmentase mixture determines general fragment length. We compared the ability of the endonucleases to generate fragments of two lengths, 100–300 bp and 300–600 bp, respectively. Following fragmentation, we prepared libraries for sequencing with the NEBNext Library Prep kit. Unique indices were ligated following end repair and dA-tailing of the fragments. Fragment size was determined by the Agilent 2100 Bioanalyzer instrument. Samples were equimolarly pooled and ran on the Illumina MiSeq using Reagent kit v2 for both 300 and 500 cycle runs. Results The two Fragmentase incubation times suggested by the manufacturer produced two very different average fragment sizes. Including the adaptor and index, the average size of the 100–300 bp incubation time was 273 bp and the 300–600 bp incubation time yielded a product of 619 bp. Conclusions Use of the NEBNext dsDNA Fragmentase allows the generation of fragments of different sizes. Although we easily generated our desired fragment sizes, we found the NEBNext DNA Library Prep Master Kit to be extremely time- and labor intensive requiring numerous transfer steps. Additionally, this kit requires a large number of AMPure XP Beads for sample clean-up. Analyses are in progress to evaluate the effect that size and library preparation method has on HLA typing assignments.

Aim To evaluate two different methods of library preparation for next generation sequencing of HLA-A, B and C on the Illumina MiSeq. Methods We initially performed long-range PCR for HLA-A, B and C on six IHWG samples using a combination of in-house and published primers. The amplified products ranged from 3900-4500 bp in length and encompassed all exons, introns and UTRs. After quantitation by the Qubit Fluorimeter (Life Technologies), samples were normalized and each sample was pooled for HLA-A, B and C. Samples were physically fragmented using sonication via the Covaris S220 instrument or enzymatically fragmented using Illumina’s Nextera XT Sample Preparation Kit. Following sonication, TruSeq adaptors were ligated and the library amplified. The Nextera kit protocol was followed as directed for the enzymatically fragmented samples. For both groups, HLA-A, B and C amplicons for each sample were pooled and indexed by sample. Fragment size was determined by either the Caliper LabChip GX or the Agilent 2100 Bioanalyzer instruments. After analysis, samples were pooled at an 8pM concentration and ran on the Illumina MiSeq using Reagent kit v2 (500 cycle). All samples were analyzed using Omixon’s HLA typing module. Results The two fragmentation methods yielded two very different average fragment sizes. The sonicated samples had an average fragment size of 488 bp while the Nextera kit samples had an average fragment size of 823 bp. Overall the shorter (sonicated) samples yielded more accurate HLA typings. However, there was one sample where a phasing issue was resolved because of the coverage provided by the longer-sized fragments. The shorter-sized sonicated samples had a greater read depth than the longer-sized Nextera kit samples. Conclusions Shorter-sized fragments appear to produce more accurate HLA typings, although, this may increase the number of phase ambiguities to be resolved. More work remains to be done to determine the most ideal fragment length size.

Aim Develop a streamlined next generation sequencing assay that enables allele level HLA typing of 16 samples at HLA-A, B, C, DRB1, DQB1 and DQA1 loci in a single Illumina MiSeq run. Methods 16 samples are amplified at 6 loci (HLA-A, B, C, DRB1, DQB1 and DQA1) by long range PCR using custom designed primers that delineate full length HLA genes (5’UTR to 3’UTR) with the exception of DRB1 for which the amplicon spans intron 1 to intron 4. Individual libraries of each amplicon are prepared by enzymatic fragmentation, end repair, adenylation and ligation of indexed adaptors. The 96 indexed libraries are pooled prior to paired-end 2 x 250 bp sequencing on an Illumina MiSeq. 10,000 paired sequence reads per library are analyzed with NGSengine (GenDx) and HLA Target (Omixon). The full workflow from amplification to HLA genotype is completed in 4 days. Sample prep is performed in a 96 well plate, enabling simple manual or automated sample preparation. Results We have successfully sequenced up to 96 libraries (16 samples x 6 loci) in a single MiSeq run. Sequence output ranged from 2.4-7.3 Gb of sequence (4.7-14 million sequence reads) per run. The quality of sequence obtained was 83.4% of bases ⩾Q30. Average depth of coverage was 461 from the alignment of 10,000 reads per locus. The accuracy of genotype results was 96.7%. Results were comparable between manual and automated library preps. Conclusions We have developed an optimized assay for the Illumina MiSeq that is a simple, robust, and scalable. The method delivers allele level HLA typing without ambiguity. Full genomic sequence is obtained for all loci with the exception of DRB1. The method can be performed manually yet is amenable to automation. Here we describe the means to HLA type 16 samples at 6 loci on the Illumina MiSeq. Given that

Aim Sequencing Based Typing (SBT) is the gold standard for high resolution typing, as the technique analyses the actual DNA sequence of the relevant HLA regions. However, Sanger SBT strategies often show ambiguous results, caused by sequencing PCR fragments with a mix of two alleles. Additional sequencing reactions are required to resolve the ambiguities. This can be overcome with Next Generation Sequencing (NGS) technology. One of the unique features of NGS is that reads originate from a single molecule, thus automatically determining the two alleles separately. This allows for development of an HLA genotyping system that yields typing results without ambiguities. In addition, NGS technology is developing fast and has come to a point where it can be applied routinely. Methods We developed NGSengine®, a software package to perform HLA typing based on NGS data. NGSengine® is an integrated approach which includes determining HLA locus, applying a sequence reference, aligning reads, phasing and typing. Since HLA is highly polymorphic, each of these steps contains HLA-specific features. In this updated study we show that HLA typing by NGS can be routinely applied using data of different NGS systems. Results Data was generated on Roche 454, Ion Torrent PGM, PacBio, Illumina MiSeq and HiSeq. All data available from these platforms could be analyzed by NGSengine®. In a single analysis the expected HLA typings were obtained. In case PCR fragments were generated with whole gene amplification (NGS-go®) no ambiguities were found since alleles were effectively separated, and the complete gene was covered by the reads. Conclusions NGSengine®: · performs reliable unambiguous HLA allele assignment based on NGS data · enables HLA-typing of NGS data in a single procedure · can be applied on data from different NGS platforms · can be applied on data generated with different amplification strategies · determines parts of allele sequences not yet present in the IMGT/HLA database · is easy to use by ‘single button analysis’. Kooter: GenDx: Employee. Ruzius: GenDx: Employee. Penning: GenDx:Employee. Mulder: GenDx: Employee; Stockholder. Rozemuller: GenDx:Employee; Stockholder.

Aim Next Generation Sequencing (NGS) has entered routine in HLA sequencing laboratories. We implemented a short amplicon-based NGS workflow for high-throughput donor registry HLA typing of 6 loci at high resolution. An elaborate primer design was key for this cost-efficient approach. Methods Potential primer binding sites are restricted by the read lengths of NGS systems. For most amplicons this requires a mix of primers that has to be optimized for uniform amplification of all alleles. In our case primers were designed to amplify under identical PCR conditions for all amplicons to facilitate the use of the 48.48 Fluidigm Access Array. Using the Illumina MiSeq 2x250bp system we chose primer binding sites to achieve full exon coverage with an overlap of 50 to 200 bases of exons 2 and 3 for HLA loci A, B, C, DRB1, DQB1 and DPB1. Self-developed Perl and Python scripts were used for analysis of each primer’s performance, as individual NGS reads can be classified and quantified. Primer sets were assessed for the ratio of target reads between amplicons and the ratio of target reads to off-target reads. Off-target reads were classified in three categories. Uniformity of amplification was assessed with the minimum requirement of a 1:5 ratio of target reads between alleles in a heterozygous sample. Based on this approach over 700 primers were screened and a final set of 66 primers was chosen. Results In three MiSeq runs 1140 samples (6840 loci) were typed and compared with previous Sanger sequencing results. All typings were in full concordance. In average 80.2% of the allocated reads corresponded to target HLA alleles, with a range from 63.9% (HLA-B exon 2) to 89.2% (HLA-A exon 2). Conclusions The validated primer set was launched in December 2012 and yields 98.7% high resolution typing. In an automated workflow, this approach allows very cost-efficient high-throughput (currently about 1500 donors per day) HLA donor registry typing.

Aim Currently applied technologies for HLA typing often show ambiguous results. For SBT these are caused by sequencing PCR fragments with a mixture of two alleles. Additional reactions are required to resolve these ambiguities. One of the unique features of Next Generation Sequencing (NGS) is that reads originate from a single molecule. Data obtained by NGS automatically determines separated results of the alleles present. This allows for the development of an HLA genotyping system that yields no ambiguities. In addition, NGS technology is developing fast and has come to a point where it can be applied routinely. We developed NGSengine, a software package to perform HLA typing based on NGS data. In this study we show that HLA typing by NGS can be routinely applied using different NGS systems. Methods NGSengine is an integrated HLA typing approach which includes determining the reference, aligning reads, phasing and typing into one method. Since HLA is highly polymorphic, each of these steps contains HLA-specific features. Data generated from various NGS platforms (Roche/454 GSFLX, Ion Torrent PGM, and Illumina MiSeq) were used to evaluate NGSengine. The sequencing data were kindly provided by D. Monos – Children’s Hospital of Philadelphia. Results All data available from different NGS platforms could be analyzed by NGSengine. In a single analysis the expected HLA typings were obtained. Furthermore results contained no ambiguity or alternative genotypes due to the complete sequencing of the entire HLA gene. Further results will be discussed in detail. Conclusions NGSengine performs reliable HLA typing based on NGS data. NGSengine enables HLA-typing of NGS data in a single procedure. NGSengine can be applied on data from different NGS platforms. Rozemuller: GenDx: Employee; Stockholder. Penning: GenDx: Employee. Mulder: GenDx: Employee; Stockholder.