26
Computer Interpretation of Uncertain DNA Evidence National Institute of Justice National Institute of Justice Computer v. Human Computer v. Human June, 2011 June, 2011 Arlington, VA Arlington, VA Mark W Perlin, PhD, MD, PhD Mark W Perlin, PhD, MD, PhD Cybergenetics, Pittsburgh, PA Cybergenetics, Pittsburgh, PA Cybergenetics © 2003-2011 Cybergenetics © 2003-2011

Computer Interpretation of Uncertain DNA Evidence National Institute of Justice Computer v. Human June, 2011 Arlington, VA Mark W Perlin, PhD, MD, PhD

Embed Size (px)

Citation preview

Page 1: Computer Interpretation of Uncertain DNA Evidence National Institute of Justice Computer v. Human June, 2011 Arlington, VA Mark W Perlin, PhD, MD, PhD

Computer Interpretation of Uncertain DNA Evidence

National Institute of JusticeNational Institute of JusticeComputer v. Human Computer v. Human

June, 2011June, 2011Arlington, VAArlington, VA

Mark W Perlin, PhD, MD, PhDMark W Perlin, PhD, MD, PhDCybergenetics, Pittsburgh, PACybergenetics, Pittsburgh, PA

Cybergenetics © 2003-2011Cybergenetics © 2003-2011

Page 2: Computer Interpretation of Uncertain DNA Evidence National Institute of Justice Computer v. Human June, 2011 Arlington, VA Mark W Perlin, PhD, MD, PhD

Uncertain DNA EvidenceEvidence

itemEvidence

data

DNA Lab Objectively Infer

10 11 12

Evidence genotype

Suspect genotype

10, 12 @ 50%11, 12 @ 30%12, 12 @ 20%

10, 12

CompareAfterwards

probability

Page 3: Computer Interpretation of Uncertain DNA Evidence National Institute of Justice Computer v. Human June, 2011 Arlington, VA Mark W Perlin, PhD, MD, PhD

Automated STR Mixture Analysis NIJ Award 2001-

IJ-CX-K003Provide quantitative DNA mixture analysis

on a wide range of data.

Handle known STR artifacts and adapt to new marker technologies.

Interact well with users, including DNA analysts and reporting officers.

Page 4: Computer Interpretation of Uncertain DNA Evidence National Institute of Justice Computer v. Human June, 2011 Arlington, VA Mark W Perlin, PhD, MD, PhD

90%90%70%70%50%50%30%30%10%10%

1, 1/2, 1/4, 1/8 ng1, 1/2, 1/4, 1/8 ng

Study Design: 40 Mixture Items

5 weights

4 dilutions

2 pairs

Margaret KlineNIST

Page 5: Computer Interpretation of Uncertain DNA Evidence National Institute of Justice Computer v. Human June, 2011 Arlington, VA Mark W Perlin, PhD, MD, PhD

MW Perlin. Real-time DNA investigation. Proceedings of Promega's 16th International Symposium on Human Identification, 2005.

Same Data, More Informationcomputer

human

info

rmat

ion

Page 6: Computer Interpretation of Uncertain DNA Evidence National Institute of Justice Computer v. Human June, 2011 Arlington, VA Mark W Perlin, PhD, MD, PhD

Computer = million x Human

MW Perlin, MM Legler, CE Spencer, JL Smith, WP Allan, JL Belrose, BW Duceman. Validating TrueAllele® DNA mixture interpretation. Journal of Forensic Sciences, 2011.

computercomputer

humanhuman

info

rmat

ion

info

rmat

ion

13.2613.26

7.037.036.246.24

10 million10 million

10 trillion10 trillion

Page 7: Computer Interpretation of Uncertain DNA Evidence National Institute of Justice Computer v. Human June, 2011 Arlington, VA Mark W Perlin, PhD, MD, PhD

Information Results

MW Perlin, A Sinelnikov. An information gap in DNA evidence interpretation. PLoS ONE, 2009.

info

rmat

ion

DNA quantity

Page 8: Computer Interpretation of Uncertain DNA Evidence National Institute of Justice Computer v. Human June, 2011 Arlington, VA Mark W Perlin, PhD, MD, PhD

Computer v. Human

MW Perlin, A Sinelnikov. An information gap in DNA evidence interpretation. PLoS ONE, 2009.

info

rmat

ion

Page 9: Computer Interpretation of Uncertain DNA Evidence National Institute of Justice Computer v. Human June, 2011 Arlington, VA Mark W Perlin, PhD, MD, PhD

Commonwealth v. Foley

Score Method13 thousand threshold

23 million use victim189 billion quantitative

• probability modeling preserves information• peak threshold discards information

The DNA Investigator™ Newsletter, 2009 Same Data, More Information – Murder, Match and DNA

homicide: 7% DNA mixture fingernail evidence

Page 10: Computer Interpretation of Uncertain DNA Evidence National Institute of Justice Computer v. Human June, 2011 Arlington, VA Mark W Perlin, PhD, MD, PhD

Predictably Preserves Information

15 loci

12 loci

67

info

rmat

ion

MW Perlin, A Sinelnikov. An information gap in DNA evidence interpretation. PLoS ONE, 2009.

Page 11: Computer Interpretation of Uncertain DNA Evidence National Institute of Justice Computer v. Human June, 2011 Arlington, VA Mark W Perlin, PhD, MD, PhD

Commonwealth v. Lyonshomicide: DNA mixture evidence

Victim's blood spatterpulsating spray from severed carotid artery

75% victim25% other

Page 12: Computer Interpretation of Uncertain DNA Evidence National Institute of Justice Computer v. Human June, 2011 Arlington, VA Mark W Perlin, PhD, MD, PhD

Degraded DNA Mixture

Degraded other DNA

Victim DNA

Page 13: Computer Interpretation of Uncertain DNA Evidence National Institute of Justice Computer v. Human June, 2011 Arlington, VA Mark W Perlin, PhD, MD, PhD

Quantitative Mixture EvidenceQuantitative peak heights at locus D8S1179

victim

other

Perlin MW, Szabady B. Linear mixture analysis: a mathematical approach to resolving mixed DNA samples. Journal of Forensic Sciences, 2001.

Page 14: Computer Interpretation of Uncertain DNA Evidence National Institute of Justice Computer v. Human June, 2011 Arlington, VA Mark W Perlin, PhD, MD, PhD

How the Computer ThinksConsiders every possible genotype solution

victimother

Explains thepeak pattern Better explanation

gives a higher probability

MW Perlin, MM Legler, CE Spencer, JL Smith, WP Allan, JL Belrose, BW Duceman. Validating TrueAllele DNA mixture interpretation. Journal of Forensic Sciences, 2011.

Page 15: Computer Interpretation of Uncertain DNA Evidence National Institute of Justice Computer v. Human June, 2011 Arlington, VA Mark W Perlin, PhD, MD, PhD

Objectively Infer Genotype

victimother

MW Perlin, MM Legler, CE Spencer, JL Smith, WP Allan, JL Belrose, BW Duceman. Validating TrueAllele DNA mixture interpretation. Journal of Forensic Sciences, 2011.

Page 16: Computer Interpretation of Uncertain DNA Evidence National Institute of Justice Computer v. Human June, 2011 Arlington, VA Mark W Perlin, PhD, MD, PhD

DNA Match Information

Probability(evidence match)

Probability(coincidental match)

How much more does the suspect match the evidencethan a random person?

6x

MW Perlin. Explaining the likelihood ratio in DNA mixture interpretation. Proceedings of Promega's 21st International Symposium on Human Identification, 2010.

LR = 6

Page 17: Computer Interpretation of Uncertain DNA Evidence National Institute of Justice Computer v. Human June, 2011 Arlington, VA Mark W Perlin, PhD, MD, PhD

Thresholds Discard Data

Over threshold,peaks becomeallele events.

Under threshold,alleles vanish.

All-or-none allele peaks; ignore victim genotype

Threshold

Page 18: Computer Interpretation of Uncertain DNA Evidence National Institute of Justice Computer v. Human June, 2011 Arlington, VA Mark W Perlin, PhD, MD, PhD

Thresholds Disperse GenotypeHuman mixture interpretation probability distribution

other

Page 19: Computer Interpretation of Uncertain DNA Evidence National Institute of Justice Computer v. Human June, 2011 Arlington, VA Mark W Perlin, PhD, MD, PhD

Thresholds Discard InformationHuman interpretation is less informative

2x

LR = 2

Page 20: Computer Interpretation of Uncertain DNA Evidence National Institute of Justice Computer v. Human June, 2011 Arlington, VA Mark W Perlin, PhD, MD, PhD

Likelihood RatioSimple Statement

A match between the suspect and the evidenceis 9.5 trillion times more probable

than coincidence.

ComputerHuman

9,500,000,000,00042,000

MW Perlin. Explaining the likelihood ratio in DNA mixture interpretation. Proceedings of Promega's 21st International Symposium on Human Identification, 2010.

Page 21: Computer Interpretation of Uncertain DNA Evidence National Institute of Justice Computer v. Human June, 2011 Arlington, VA Mark W Perlin, PhD, MD, PhD

Missed Allele Error > 100%

The DNA Investigator™ Newsletter, 2011 DNA Intelligence and Forensic Failure – What you don't know can kill you

Page 22: Computer Interpretation of Uncertain DNA Evidence National Institute of Justice Computer v. Human June, 2011 Arlington, VA Mark W Perlin, PhD, MD, PhD

Human Information

25caseitems

inconclusive

Page 23: Computer Interpretation of Uncertain DNA Evidence National Institute of Justice Computer v. Human June, 2011 Arlington, VA Mark W Perlin, PhD, MD, PhD

Computer Information

25+27

more

resolved

Page 24: Computer Interpretation of Uncertain DNA Evidence National Institute of Justice Computer v. Human June, 2011 Arlington, VA Mark W Perlin, PhD, MD, PhD

SWGDAM TriageInterpret DNA evidence

• Library of match information• On-demand case report

humanstochasticthreshold

computerprobabilistic

genotype(3.2.2)

resolvable inconclusive

Page 25: Computer Interpretation of Uncertain DNA Evidence National Institute of Justice Computer v. Human June, 2011 Arlington, VA Mark W Perlin, PhD, MD, PhD

Computer v. HumanNo competition

• information• reproducible• mixture resolution• degraded DNA• unsolved cases• complex cases• interpretation• likelihood ratio• crime impact

1,000,000xValidated1, 2, 3, 4, …QuantifiedSolvedAutomatedEasy & fastExplainedMaximized

People use computer tools to do a better job

Page 26: Computer Interpretation of Uncertain DNA Evidence National Institute of Justice Computer v. Human June, 2011 Arlington, VA Mark W Perlin, PhD, MD, PhD

Learning More

www.cybgen.com/information

• Newsletters gentle introduction to ideas• Courses for scientists and lawyers• Presentations handouts, movies, transcripts• Publications abstracts, manuscripts

The science of quantitative DNA mixture interpretation