Applications of NGS to Forensic mtdna Analysis

Transcription

1 Mitchell M. Holland, Ph.D. Associate Professor, Biochem & MolBio Former Director, Forensic Science Program Penn State University, University Park, PA Applications of NGS to Forensic mtdna Analysis MN BCA Workshop 1 June

2 Topics to Address Why NGS for forensic mtdna analysis? Preliminary studies: Detection of low-level heteroplasmy Recent findings: Mutation & heteroplasmy rates, bottlenecks, transmission of heteroplasmy Ongoing Research

3 Is this the best target for introducing NGS into crime labs & the legal system?? mtdna Sequencing

4 Historical Reference RFLP ~1987-Late 1990 s The forensic community DIDN T go directly from RFLP to STRs STRs ~1991-Present

5 Historical Reference RFLP ~1987 to Late 1990 s Fast Track DQA1/PM ~1990 to Late 1990 s Slower Track STRs ~1991 to Present

6 Historical Reference DQA1/PM testing provided an easier path for the admissibility of PCR in courts of law paving the way for STRs

7 The Future?? STRs Capillary Electrophoresis Are we going directly from STRs by CE to both STRs & SNPs by NGS? STRs & SNPs Next Generation Sequencing

8 or will history repeat itself?? mtdna testing may provide a similar path for the admissibility of NGS in courts of law paving the way for STRs and SNPs

9 Search of NGS + mtdna = >475 journal articles

10 Search of NGS + mtdna = >475 journal articles Methods in Molecular Biology book with multiple chapters on NGS of mtdna 2015

11 Search of NGS + mtdna + forensic = 27 journal articles

12 Why bother changing from Sanger sequencing to NGS for forensic mtdna analysis?

13 Why bother changing from Sanger sequencing to NGS for forensic mtdna analysis? Heteroplasmy Detecting & Reporting

14 Identification of Nicholas Romanov Tsar Nicholas II Family Reference 5 Generations Removed

15 DGGE to Identify Heteroplasmy Family Reference 5 Generations Removed Used DGGE analysis to identify the heteroplasmic sequences including from the distant maternal relative

16 Identification of Nicholas Romanov Tsar Nicholas II Weight of the Evidence Georgij Romanov LR = ~150 LR = ~380,000 When the heteroplasmy was considered

17 Likelihood Ratios LR = p(e1/r) x p(e2/r) p(e1/r ) x p(e2/r ) p(e1/r) = the probability of the evidence (match between Georgij and Nicholas) given the hypothesis that the remains are those of Nicholas Romanov E2 = the probability of co-occurrence of heteroplasmy R = given the hypothesis that the remains are unrelated

18 Likelihood Ratios Likelihood of a Match LR = p(e1/r) p(e1/r ) p(e1/r) = e -g = 0.96 g = number of generations between individuals = 2 m = mutation rate, which we estimated as 1/50 generations Remember: This value would be 1 (one) when matching forensic evidence to an individual

19 Likelihood Ratios LR = p(e1/r) p(e1/r ) = 0.96/ = 148 Rounded to 150 p(e1/r) = e -g = 0.96 g = number of generations between individuals = 2 m = mutation rate, which we estimated as 1/50 generations P(E1/R ) = the haplotype frequency of 2/308 = one observation in a database of 307 unrelated people, and one observation in the case = Remember: When matching forensic evidence to an individual, the value would be 1/ = 154 EMPOP now has ~35,000 sequences in the database, f = (upper bound of 95% CI)

20 Likelihood Ratios Likelihood of Sharing Heteroplasmy LR = p(e2/r) p(e2/r ) p(e2/r) = e -g = 0.61 g = number of generations between individuals = 2 = rate of fixation to (apparent) homoplasmy = ¼ for the Romanov lineage (fixation in 4 generations) = 0.25 Remember: This value would be 1 (one) when matching forensic evidence to an individual

21 Likelihood Ratios When matching forensic evidence to an individual, the value would be 1/2.4X10-4 = 4.2X10 3 LR = p(e2/r) p(e2/r ) = 0.61/2.4 X 10-4 = 2.5 X 10 3 p(e2/r) = e -g = 0.61 g = number of generations between individuals = 2 = rate of fixation to (apparent) homoplasmy = ¼ for the Romanov lineage (fixation in 4 generations) = 0.25 P(E2/R ) = the chance of randomly sampling another individual with heteroplasmy = 4/50 X 1/332 = 2.4 X /50 = the chance of randomly selecting individuals with heteroplasmy 1/332 = polymorphic sites identified in HV1/HV2 (332/610)

22 Likelihood Ratios LR = p(e1/r) x p(e2/r) p(e1/r ) x p(e2/r ) When matching forensic evidence to an individual, the value would be (154)(4.2X10 3 ) = 6.5X10 5 = (150)(2.5 X 10 3 ) = 3.8 X 10 5 Therefore, the DNA evidence is 380,000 times more likely if the remains are Nicholas Romanov This is a conservative estimate!!

23 Lays the Framework for What is the true mutation rate of mtdna? How are heteroplasmic variants transmitted from one generation to the next? What is the rate of heteroplasmy? Within population groups? On a per nucleotide basis? 2014-DN-BX-K022

24 How are the questions related? Mutations lead to heteroplasmic states Mutations in the germline are transmitted to the next generation At what point does the sequence become homoplasmic? Mutation or Reversion? Regeneration?

25 Bottleneck At what point does the sequence become homoplasmic? Mutation or Reversion?

26 Substitution Rate of mtdna 1997 We compared DNA sequences of the two CR hyper-variable segments from close maternal relatives, from 134 independent mtdna lineages spanning 327 generational events. Ten substitutions were observed, resulting in an empirical rate of 1/33 generations, or 2.5/site/Myr. This is roughly twenty-fold higher than estimates derived from pylogenetic studies; /site/myr.

27 Substitution Rate of mtdna 1997 Ten substitutions were observed, resulting in an empirical rate of 1/33 generations, or 2.5/site/Myr. This is roughly twenty-fold higher than estimates derived from pylogenetic studies; /site/myr. Changes the initial LR to 0.94/ = 145 and to 2.4X10 3 overall for the ID of the Tsar BUT, is 1/33 a true mutation rate?

28 Substitution Rate of mtdna 1997 Using our empirical rate to calibrate the mtdna molecular clock would result in an age for the mtdna MRCA of only ~6,500 y.a., clearly incompatible with the known age of modern humans. The mtdna substitution rate is sufficiently high that differences between true maternal relatives will be encountered frequently, providing the grounds for false exclusions. Therefore, it is our policy not to report an exclusion based on a single CR sequence difference.

29 Substitution Rate of mtdna 1997 Heteroplasmy was clearly detected in five individuals from three lineages. While this adds to a growing body of evidence that point mutation heteroplasmy in the human CR may not be particularly rare, a suitable estimate of the frequency of heteroplasmy is still lacking. This is because direct (Sanger) sequencing is not an appropriate method for determining the frequency of heteroplasmy, as even fairly balanced heteroplasmic ratios can be difficult to distinguish from various types of sequencing artifacts.

30 High Rate of Heteroplasmy in HV1

31 Enter stage right next generation DNA sequencing (NGS)

32 Our Initial NGS Work: (Under Mitch Holland s Research Page) Croat Med J (2011), 52, pp

33 Initial Considerations Concordance With Sanger Data Initial Rate of Heteroplasmy Repeatability (same operator & instrument) Threshold Definitions/Interpretation Criteria Error Rates

34 Evaluated 30 individuals from 25 different mtdna lineages HV1 Data Concordance Sample F2 F3 Sanger mtdna Profile 16069T, 16093C, 16126C, 16261T, 16274A, 16355T 16069T, 16126C, 16145A, 16172C, 16261T Percent of Minor Heteroplasmy & Site % C Not Detected 454 GS Junior mtdna Profile 16069T, 16093C, 16126C, 16261T, 16274A, 16355T 16069T, 16126C, 16145A, 16172C, 16261T Percent of Minor Heteroplasmy & Site % T % C % C Not Detected F4 No polymorphisms Not Detected No polymorphisms Not Detected F5 F7, F12-13, M A, 16172C, 16223T, 16311C 16192T, 16256T, 16270T Not Detected Not Detected 16129A, 16172C, 16223T, 16311C 16192T, 16256T, 16270T % G % T % C F T, 16362C Not Detected 16223T, 16362C % C F C Not Detected 16356C Not Detected F C Not Detected 16298C % T F C, 16239T, 16294T, 16296T, 16304C Not Detected Table 3, Holland et al, CMJ C, 16239T, 16294T, 16296T, 16304C Not Detected F G Not Detected 16343G Not Detected F C Not Detected 16093C Not Detected F C, 16278T Not Detected 16172C, 16278T Not Detected M T Not Detected 16355T Not Detected M T Not Detected 16111T % C 0.33% or 1/300

35 Sanger versus NGS Heteroplasmy Detection Figure 2, Holland et al, CMJ 2011 SANGER NGS 3.71% C/T Heteroplasmy 1.29% T/C Heteroplasmy 20.14% C/T Heteroplasmy

36 Evaluated 30 individuals from 25 different mtdna lineages Sample Sanger mtdna Profile Percent of Minor Heteroplasmy & Site 454 GS Junior mtdna Profile Percent of Minor Heteroplasmy & Site 11/25 (44%) of the lineages showed some level of heteroplasmy With 19 sites of heteroplasmy 32% (8/11) of the lineages with 1%+ heteroplasmy M5 M A, 16129A, 16192T, 16213A, 16223T, 16278T, 16355T, 16362C 16129A, 16223T, 16264T Not Detected Not Detected 16114A, 16129A, 16192T, 16213A, 16223T, 16278T, 16355T, 16362C 16129A, 16223T, 16264T % C Not Detected M C, 16311C Not Detected 16224C, 16311C Not Detected M T, 16343G, 16356C Not Detected 16301T, 16343G, 16356C Not Detected M C Not Detected 16304C % C % T % T M A, 16223T Not Detected 16129A, 16223T Not Detected M T, 16126C Not Detected 16069T, 16126C % T M15 M17 M18 M19, F C, 16224C, 16311C 16126C, 16294T, 16296T 16278T, 16304C, 16311C 16069T, 16126C, 16222T Not Detected Not Detected Not Detected Not Detected 16093C, 16224C, 16311C 16126C, 16294T, 16296T 16278T, 16304C, 16311C 16069T, 16126C, 16222T % T Not Detected % T % C % G % T Not Detected

37 Other Examples 64 mtgenomes <0.02% Differences from Sanger Data Most Differences in Homopolymeric Stretches PGM capable of producing quality, reliable mtdna sequence data Concordance

38 Repeatability Sample Sanger mtdna Profile Percent of Minor Heteroplasmy & Site 454 GS Junior mtdna Profile Percent of Minor Heteroplasmy & Site Is low level heteroplasmy reproducible? M5 M A, 16129A, 16192T, 16213A, 16223T, 16278T, 16355T, 16362C 16129A, 16223T, 16264T Not Detected Not Detected 16114A, 16129A, 16192T, 16213A, 16223T, 16278T, 16355T, 16362C 16129A, 16223T, 16264T % C Not Detected M C, 16311C Not Detected 16224C, 16311C Not Detected M T, 16343G, 16356C Not Detected 16301T, 16343G, 16356C Not Detected M C Not Detected 16304C % C % T % T M A, 16223T Not Detected 16129A, 16223T Not Detected M T, 16126C Not Detected 16069T, 16126C % T M C, 16224C, 16311C Not Detected 16093C, 16224C, 16311C % T M C, 16294T, 16296T Not Detected 16126C, 16294T, 16296T Not Detected M T, 16304C, 16311C Not Detected 16278T, 16304C, 16311C % T % C % G % T Repeatability M19, F T, 16126C, 16222T Not Detected 16069T, 16126C, 16222T Not Detected

39 Repeatability Sample Sanger mtdna Profile Percent of Minor Heteroplasmy & Site 454 GS Junior mtdna Profile Percent of Minor Heteroplasmy & Site M C Not Detected 16304C % C % T % T M10 Replicate # % C % T % T M10 Replicate # % C % T % T

40 Repeatability Sample Sanger mtdna Profile Percent of Minor Heteroplasmy & Site 454 GS Junior mtdna Profile Percent of Minor Heteroplasmy & Site F A, 16172C, 16223T, 16311C Not Detected 16129A, 16172C, 16223T, 16311C % G % T F5 Replicate # % G % T F5 Replicate # % G % T

41 Interpretation Criteria What is the empirical rate of error? Can thresholds or filters be applied? What effects do these answers have on the reporting of low-level mtdna heteroplasmy?

42 Reporting Thresholds Poisson Distribution = Predicts the degree of spread around a known average # of Minor Site Observations 95% of the error is below this level Poisson Distribution with (mean) Corresponding to the AT Reporting (Interpretation) Threshold (RT) Analytical Threshold (AT) Nucleotide Position

43 Reporting Thresholds Empirical Cumulative Distribution Function Maximum, Tiered, or Rolling Approach??

44 Reporting Thresholds

45 Recent Studies (MiSeq) Whole mtgenome sequencing on 160+ individuals Some of the data was replicated Expanded assessment of low level heteroplasmy rates Included in the data set were 50 maternal pairs to assess inheritance patterns of the low level variants (blind to us) including tissue specific differences Mutation rates and bottleneck assessments

46 Rate of Heteroplasmy Data Set = 109 Individual Lineages (50 Pairs of Maternal Relatives) % Heteroplasmy >1% Heteroplasmy >10% Heteroplasmy Coding Region 69% 50% 14% Control Region 50% 26% 8.6%* *Consistent with previous reports: for example, Irwin et al, J Mol Evol 2009

47 Sites of Heteroplasmy >1-10% Heteroplasmy C/T 73 G/A C/T 143 A/G C/T (twice) 185 A/G (3 times) A/G (twice) 214 A/G T/C 223 T/C A/G 225 A/G T/C C/T A/G (twice) T/C C/T T/C C/T G/A A/G G/A G/A T/C G/A T/C >10% Heteroplasmy A/G T/C (twice) C/T C/T

48 Genetic Bottlenecks & Empirical Mutation Rates Given our data, the number of mtdna molecules transmitted to the next generation is Given our data, the germline mutation rate is 0.13 mutations/site/myr (compared to phylogenic rate estimates of 0.118) Nonsynonymous mutations have gone through a purifying selection process Proceedings of the National Academy of Sciences

49 Maternal Age Effects No age association was found for the children, but there was a correlation between age of the mothers & rate of heteroplasmy, and a correlation between age of the mothers at fertilization & rate of heteroplasmy in the children

50 Age Effects on Tissue- Associated Heteroplasmy We expect heteroplasmy allele frequency to diverge less in the tissues of a child than in those of a mother, which is what we observed The frequencies were more strongly correlated between the two tissues for children (R 2 = 92%) than between the two tissues for mothers (R 2 = 49%) Hematopoietic Stem Cells (derived from mesoderm in the early embryo) versus Epithelial Cells (stratified squamous)

51 Age Effects on Tissue- Associated Heteroplasmy The lack of correlation between the tissue types of mothers and children is a result of the action of the germ-line bottleneck

52 Maternal Transmission 1 of 50 maternal pairs (mother:child) had no sites of heteroplasmy and matched completely 49 of 50 pairs could be differentiated through heteroplasmic differences 24 sites of heteroplasmy 11/24 sites NOT shared, 3/11 in the CR (15%) 215, and Maternal Transmission Control Region = CR

53 Site Differences Between Maternal Relatives #1098 #1100 Primary Haplotype A263G Heteroplasmy Positions 200 A/G (3.0%) Primary Haplotype A263G Heteroplasmy Positions T16093C 16093T/C (12.6%) C16261T C16291C T16311C T16362C T16519C T16093C 16093C/T (3.4%) C16261T C16291C T16311C T16362C T16519C

54 Site Differences Between Maternal Relatives #618 #606 Primary Haplotype A73G A263G Heteroplasmy Positions Primary Haplotype A73G A263G Heteroplasmy Positions T16093C 16093C/T (11.4%) T16224C T16311C T16519C T16093C T16224C T16311C T16519C

55 Site Differences Between Maternal Relatives #1161 Primary Haplotype G203A T204C T239C A263G C16193T A16219G T16362C A16482G Heteroplasmy Positions A16037A/G (44%) G16390G/A (0.9%) #1279 Primary Haplotype G203A T204C T239C A263G Heteroplasmy Positions C150C/T (0.58%) A16037G/A (3.0%) C16193T C16193T/C (0.53%) A16219G T16362C A16482G

56 Inheritance Patterns 1 2 Sisters 3 4 Cousins

57 Inheritance Patterns Tissue Variant Drift Nucleotide Position G185AG Buccal Buccal Blood (3) or Duplicate PCR (1, 2, 4) % 2.49% 2.41% % 2.50% 2.95% % 1.94% 0.89% % 2.49% 1.73%

58 Inheritance Patterns Tissue Variant Drift Nucleotide Position A189AG Buccal Buccal Blood (3) or Duplicate PCR (1, 2, 4) % 1.31% 1.40% % 1.62% 1.39% % 1.16% 0.54% % 1.24% 0.91%

59 Inheritance Patterns Tissue Differences Nucleotide Position A215AG Buccal Buccal Blood (3) or Duplicate PCR (1, 2, 4) 1 <0.5% <0.5% <0.5% 2 <0.5% <0.5% <0.5% % 1.30% <0.5% 4 <0.5% <0.5% <0.5%

60 Current Studies NIJ 2014-DN-BX-K022 Rate of heteroplasmy in the entire D-loop for ~600 unrelated individuals Individuals of European origin Three age ranges: 18-30, & 51+ Approximately 50:50 male:female Rates on a per nucleotide basis Transmission: between generations & tissue types Provide recommendations for best practices

61 Current Studies Validation of the D-loop protocol from illumina Optimize 1 st round amplification, assess sensitivity, repeatability, mixtures, etc, for the MiSeq process The impact of DNA damage on NGS results from the entire D-loop Passive, active, low-level template

62 Thanks!! Illumina MiSeq Cydne Holt, Kathy Stephens, Joe Valaro, Carey Davis, Dan Gheba, etc Liam Phillips Katie O Hanlon AIBiotech & Megan Connolly OCSO, MI Saransh Kabra PSU Master s Student in Biotechnology Manfred Kayser Erasmus MC, Netherlands Cedric Neumann South Dakota State University SoftGenetics NextGENe John Fosnacht, Teresa Snyder-Leiby, etc 454 LifeSciences/Roche GS Junior Walther Parson & Ann Gross

63 Penn State Collaborators Kateryna Makova Collection of Samples from ~350 Families Blood, Buccal and Hair Department of Biochemistry & Molecular Biology Eberly College of Science Anton Nekrutenko

64 Battelle Collaborators Christine Baker, Charlottesville Operations Seth Faith, Columbus Operations (currently at NCSU) Brian Young, Columbus Operations FY13 Variomics IR&D Task 5.3

65 Thanks!! Jen McElhoe, Research Associate Master s Students: Molly Rathbun (damage) Laura Wilson (D-loop val) UG Students: Jaclyn Junod (enzymes) Current Research Group Alyssa Duffy Lindsay Domdrosky Jillian Baker Sean Lynch