Evaluation of MIA FORA NGS HLA test and software. Lisa Creary, PhD Department of Pathology Stanford Blood Center Research & Development Group

Transcription

1 Evaluation of MIA FORA NGS HLA test and software Lisa Creary, PhD Department of Pathology Stanford Blood Center Research & Development Group

2 Disclosure Alpha and Beta Studies Sirona Genomics Reagents, *Equipment, and Software supplied by Sirona Genomics *Thermal cyclers, Biomek 4000, and Illumina MiSeq

3 NGS-HLA typing requirements for the Stanford Blood Center Unambiguous phased genotypes Automated methods Accurate unedited genotype calls Easy to perform Easy intuitive software Cost effective No reflexive testing 3-4 day TAT

4 MIA FORA NGS HLA TECHNOLOGY Comprehensive coverage of class I (HLA-A, -B, -C) and class II (DPA, -DPB, -DQA, -DQB, -DRB1, -DRB3, - DRB4, -DRB5) genes using long range PCR Multiplexed x24 samples per Illumina MiSeq run Unambiguous phased genotypes at the 4-field level of resolution in a single pass Detection of novel and null alleles Automated and Manual methods developed

5 MIA FORA NGS HLA typing solution Low Throughput kit: 24 samples PCR Master mixes x 1 box, 9 tubes Post PCR x 1 box, 12 tubes x 1 96-well plate (indexes) Compact Easy Storage

6 Alpha Study Conducted from March 2015 to May 2015 On-site training by Sirona Genomics Biomek 4000 programs written and installed by Sirona Genomics Evaluated control DNA samples (x12) supplied by Sirona Genomics using both manual and automated methods Evaluated 138 Stanford Blood Center DNA samples x 2 runs using the manual protocol x 4 runs using the automated protocol

7 Liquid handling workstations Pre-PCR Post-PCR Perkin Elmer NGS Express Biomek 4000

8 Long Range Amplification , 5.6 Single PCR condition used for all loci Optimized extensively to preserve allele balance and prevent allele dropout

9 Pool Sirona Workflow Day 1 (~5 hours) Long Range PCR Samples 1-8 Samples 9-16 Samples Quantification, Balancing and Pooling amplicons Day 2 3 Library prep Enzymatic Fragmentation, End repair, A tailing Index adaptor ligation Pooling, Size Selection, qpcr Day 4 (24 hours) Sequencing Day 5 (4 hours) HLA genotype Assignment 2 x 150 bp paired end reads

10 Alpha study Concordance results of Sirona Control DNA samples Total alleles N = 216 A B C DPA1 DPB1 DQA1 DQB1 DRB1 DRB3 DRB4 DRB5 Manual 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% Automated 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% Unambiguous typing at 2-field level of resolution using both manual and automated methods

11 Alpha Study Stanford Blood Center Samples Unedited Genotype Calls Overall Concordance 99.1% Edited Genotype Calls Overall Concordance 99.9% Identical Identical 25.3 Ambiguous Reference, Unambiguous NGS 25.4 Ambiguous Reference, Unambiguous NGS 73.0 Ambiguous Reference, Ambiguous NGS 73.6 Ambiguous Reference, Ambiguous NGS Unambiguous Reference, Ambiguous NGS Unambiguous Reference, Ambiguous NGS Total Samples Alleles with SBT/SSP/SSO reference genotypes Number of NGS alleles

12 Alpha Study Stanford Blood Center Samples Unedited genotypes Edited genotypes A B C DPA1 DPB1 DQA1 DQB1 DRB1 DRB3 DRB4 DRB5 0 A B C DPA1 DPB1 DQA1 DQB1 DRB1 DRB3 DRB4 DRB5

13 Alpha Study Sources of discordance (Edited Results) NGS Novel alleles (exon variants) BS-25: DQA1*05:05:01:01_Exon 3 Variant, insertion of T, codon 135, frameshift mutation. S-70: DQB1*05:01:01:01_Exon 1 Variant, C to T, Ser to Ser, codon -6. S-100: Novel allele by SBT reported as C*03@/C*14 hybrid. Novel allele by NGS is hybrid of C*07:02:01:01 exon 1-intron2/C*14:02:01 exon 3-3 UTR.

14 Improvements at end of Alpha study Laboratory work Increased pooling amount of DRB amplicons to improve balancing Software database Cloned and sequenced all alleles of IHWG 50 cell lines Included reference sequences for all DPA1 and DPB1 alleles Genotype calling algorithms enhanced to improve accuracy of automatic calls

15 Beta Study Conducted from August 2015 to October 2015 Evaluated 69 Stanford Blood Center DNA samples x 3 runs using the automated protocol

16 Beta Study Stanford Blood Center Samples Unedited Genotype Calls Overall Concordance 99.3% Edited Genotype Calls Overall Concordance 99.5% Identical Ambiguous Reference, Unambiguous NGS Ambiguous Reference, Ambiguous NGS Unambiguous Reference, Ambiguous NGS Identical Ambiguous Reference, Unambiguous NGS Ambiguous Reference, Ambiguous NGS Unambiguous Reference, Ambiguous NGS Disconcordant Disconcordant Total Samples Alleles with SBT/SSP/SSO reference genotypes Number of NGS alleles

17 Beta Study Stanford Blood Center Samples Unedited genotypes Edited genotypes A B C DPA1 DPB1 DQA1 DQB1 DRB1 DRB3 DRB4 DRB5 0 A B C DPA1 DPB1 DQA1 DQB1 DRB1 DRB3 DRB4 DRB5

18 Beta Study Reasons for discordance (Edited Results) NGS Novel alleles (exon variants) BS-72: A*23:01:01_Exon 4 Variant, C to T, Ala to Val, codon 211. SBC-200: B*40:02:01e1_Exon 3 Variant, T to G, Leu to Arg, codon 95. SBC-202: DQB1*05:01:01:01_Exon 1 Variant, C to T, Ser to Ser, codon -6. SBC-207: DPA1*02:02:02e1_Exon 3 Variant, C to G, Pro to Ala, codon 96. SBC-223: DQB1*05:01:01:01_Exon 1 Variant, C to T, Ser to Ser, codon -6.

19 MIA FORA NGS Software Accurate phase defined unambiguous HLA genotypes Three complementary algorithms: 1) Expected-Maximization to rank alleles based on coverage metrics: Coverage calculated from competitive alignment of PE reads with HLA reference sequences from the IMGT/HLA database and in-house reference sequences 2) Phasing assembly algorithm 3) Bayesian consensus algorithm High Performance server, multiple user access

20 MIA FORA NGS Software

21 MIA FORA NGS Software

22 MIA FORA NGS Software Confidence >=90 Confidence >=70 Confidence >=50 Confidence >=0 non-cwd CWD Overwritten Automatic Call Not found in LD DB Found in LD DB 1/2 1/2 Number of contigs different from that of alleles Number of contigs same as that of alleles? Inconsistence between genotype derived from phased and EM result.

23 SBT/SSO vs NGS Resolving an allele ambiguity Reference genotype: A*23:01/A*23:17, A*29:02 NGS genotype: A*23:17, A*29:02 A*23:01:01 x1 M/M in exon 5 S-225

24 SBT/SSO vs NGS Resolving a phase ambiguity and identifying a novel allele BS-72 Reference genotype: A*03:78, A*23:01 A*03:78 / A*03:01:01:01 1 nt difference in Exon 4

25 SBT/SSO vs NGS Resolving a phase ambiguity and identifying a novel allele BS-72, Reference genotype: A*03:78/A*03:01:01:01, A*23:01 NGS genotype: A*03:01:01:01, A*23:01:01_Exon4 variant, C > T, Ala >Val, codon 211 A*23:01:01 x1 M/M in exon 4

26 SBT/SSO vs NGS Identifying a novel allele S-101, Reference Type Result: B*13@, B*38@, one allele is an exon 4 variant? NGS Result: B*13:02:01, B*38:02:01 _Exon 4 variant A to G, Lys to Arg, codon 186.

27 DRB4*01:03:01:02Ne1 S-214 Genomic reference sequence

28 S-16 Evidence of a second allele Unmapped reads that do not align to DRB1*01:03:01:01 Second allele: DRB1*01:01:01:01

29 Evidence of a second allele S-36 Unmapped reads that do not align to DPA1*01:03:01:04 Second allele: DPA1*02:01:01_x2 intron 1 variants

30 Summary MIA FORA workflows (manual and automated) are very easy to follow Pooling amplicons early on the process reduces hands on time, decrease costs, minimizes deck space used on automated workstations High concordance (99.9%) with reference genotypes Wide gene coverage using 2 x 150 bp paired end reads minimized phased ambiguities for all loci except for some DPB1 allele combinations MIA FORA NGS able to distinguish all genotypes at 3-fields and for the majority of genotypes at 4-field level resolution Software continually improved during Alpha and Beta trials to improve accuracy of automatic calls

31 Acknowledgments Stanford Blood Center Prof Dolly Tyan Prof Marcelo Fernandez-Vina Sridevi Gangavarapu Sirona Genomics Michael Mindrinos Sujatha Krishnakumar Chunlin Wang Ming Li Tommy Liu Jennifer Simonovich Raquel Kuehn Marilyn Fukushima Farbod Babrzadeh