Cold Hits and Mixtures
Cold Hits and Mixtures This project was supported by Award No. 2010-DN-BX-K259 awarded by the National Institute of Justice, Office of Justice Programs, U.S. Department of Justice.
Basic terminology: Genetics • DNA Polymorphism (“many forms”) – Regions of DNA which differ from person to person
• Locus (plural = loci) – Site or location on a chromosome
• Allele – Different variants which can exist at a locus
• DNA Profile – The combination of alleles for an individual
Basic terminology: Technology • Amplification or PCR (Polymerase Chain Reaction) – A technique for ‘replicating’ DNA in the laboratory (‘molecular Xeroxing’)
PCR Process 5’
3’
Starting DNA 5’ Template
3’
Forward primer
3’
Separate strands (denature) Add primers 5’ (anneal)
Make copies (extend primers)
Repeat Cycle, Copying DNA Exponentially
5’
3’
Reverse primer
STR
• Short tandem repeat • Is not a test • Describes a type of DNA polymorphism in which: – a DNA sequence repeats – over and over again – and has a short (4 base pair) repeat unit
• A length polymorphism -- alleles differ in their length 3 4 5 6
repeats: repeats: repeats: repeats:
AATG AATG AATG AATG
AATG AATG AATG AATG
AATG AATG AATG AATG AATG AATG AATG AATG AATG AATG
Applied Biosytems (ABI) Automated STR Tests • Profiler Plus and Cofiler Tests – – – –
Profiler Plus tests 9 STR loci plus amelogenin Cofiler tests 6 STR loci plus amelogenin 2 loci common between two tests Often performed in tandem -- test all 13 CODIS loci
• Identifiler – Tests the 13 “core” CODIS loci plus two more
• Same 13 loci used by all labs • Tests measure the length of a piece of DNA
Basic terminology: Statisics • Random Match Probability: the probability that an randomly selected person from a given population of unrelated individuals would have the same profile as the evidence
• Combined Probability of Inclusion: The probability that a random person would be included in the observed mixture
Cold Hit • Why are cold hit cases statistically different from “probable cause” cases? Search through the databank gives the government thousands of opportunities to find a “match” The more profiles searched, the greater the chance of identifying an innocent person Databank has relatives and mixed racial groups
Database Searches and Cold Hits • Ten million tickets are sold for a lottery. What are your chances of winning if you buy: – One ticket? (1 in 10 million) – Ten tickets? (1 in 1 million) – 10,000 tickets? (1 in 1000)
• Among innocent people, 1 in 10 million has a DNA profile that would “match” the evidence. What are the chances the government will find a match if it compares the profiles of: – – – –
One innocent person? (1 in 10 million) Ten innocent people? (1 in 1 million) 10,000 innocent people? (1 in 1000) Five million innocent people? (1 in 2)
Cold Hit Statistics
• NRC I—test additional loci and report F for those loci only – Presumes ascertainment bias is a serious problem
• NRC II—report FxN, where N is the number of profiles in the database – E.g., if F=1 in 1 billion; N=1 million; then tell jury RMP=1 in 1000
• prosecutors and lab techs everywhere— ascertainment bias is not a problem, so just tell the jury F
Example of Database Match Probability NRC II—report FxN, where N is the number of profiles in the database
PEOPLE v. JOHN PUCKETT
RANDOM MATCH PROBABILITY:
1 in 1.1 million
DATABASE MATCH PROBABILITY: 1 IN 3
DATABASE MATCH PROBABILITY OR NRC II • THE KEY TO THIS STATISTIC IS THAT THE NUMBER GENERATED IS THE PROBABILITY OF FINDING A MATCH IN A DATABASE OF A GIVEN SIZE WHEN THE ACTUAL PERPETRATOR IS NOT PRESENT IN THE DATABASE
MIXTURES Subjective Interpretation
Mixture Interpretation • As the number of contributor’s grow, the data becomes harder to evaluate • Major versus minor contributors • Peak height ratios • Dropout • Dropin (LCN)
Mixture Interpretation ASSUMPTIONS • Number of contributors • Intimate samples – subtracting victim profile
“RESOLVED” OR “DEDUCED” PROFILE STATISTICS
Casework mixtures can be quite complicated • Example shows at least 4 donors • Other contributors unknown • Peak heights meaningless • Becomes difficult to exclude anyone
Mixtures
• More than two alleles at a locus may indicate a mixture • Number of contributors often unclear because of sharing alleles • Some labs rely on ‘peak height ratio’ • May be arbitrary: factors other than the quantity of DNA can effect peak height • Statistics used in mixture cases: may make debatable assumptions
Allelic Dropout D5S818
D13S817
D7S820
Evidence sample
Reference sample
• Peaks in evidence samples all very low – Mostly below 150 rfu
• At D13S817: – Evidence sample: 8, 8 – Reference sample: 8, 14
• 14 allele has dropped out • Tend to see with ‘marginal samples’
• Conclusion: Suspect included, But Why?
Allelic Dropout • Suspect Included Why? – – – –
We know allelic dropout occurs We can’t say it didn’t happen here (But can’t say it did) To be “conservative” we conclude that this is an inclusion
• Subjectivity – What is a peak, what’s not
• Examiner Bias – Included at all the other loci, therefore it must be him
A forensic lab assigned the major/minor alleles incorrectly @ THO1: major profile=6,6 and minor profile=8,9.3. Correct call is major profile=6,8 and minor profile=6,9.3 of 2-person mixture, 3:1 ratio.
Mixtures and Cold Hits
SAME SEARCH DIFFERENT RESULTS
EXAMINER BIAS AND MIXTURES SUBJECTIVITIY AND BIAS IN FORENSIC DNA MIXTURE INTERPRETATION, IN PRESS, SCIENCE AND JUSTICE ITIEL DROR AND GREG HAMPIKIAN
TASK: 17 ANALYSTS FROM SAME LAB EXAMINED THE SAME MIXTURE AND REFERENCE SAMPLES RESULT: 1 ANALYSTS INCLUDED SUSPECT 4 ANALYSTS CONCLUDED “INCONCLUSIVE” AND 12 ANALYSTS EXCLUDED THE SUSPECT
Basic terminology: Genetics • DNA Polymorphism (“many forms”) – Regions of DNA which differ from person to person
• Locus (plural = loci) – Site or location on a chromosome
• Allele – Different variants which can exist at a locus
• DNA Profile – The combination of alleles for an individual
Basic terminology: Technology • Amplification or PCR (Polymerase Chain Reaction) – A technique for ‘replicating’ DNA in the laboratory (‘molecular Xeroxing’)
PCR Process 5’
3’
Starting DNA 5’ Template
3’
Forward primer
3’
Separate strands (denature) Add primers 5’ (anneal)
Make copies (extend primers)
Repeat Cycle, Copying DNA Exponentially
5’
3’
Reverse primer
STR
• Short tandem repeat • Is not a test • Describes a type of DNA polymorphism in which: – a DNA sequence repeats – over and over again – and has a short (4 base pair) repeat unit
• A length polymorphism -- alleles differ in their length 3 4 5 6
repeats: repeats: repeats: repeats:
AATG AATG AATG AATG
AATG AATG AATG AATG
AATG AATG AATG AATG AATG AATG AATG AATG AATG AATG
Applied Biosytems (ABI) Automated STR Tests • Profiler Plus and Cofiler Tests – – – –
Profiler Plus tests 9 STR loci plus amelogenin Cofiler tests 6 STR loci plus amelogenin 2 loci common between two tests Often performed in tandem -- test all 13 CODIS loci
• Identifiler – Tests the 13 “core” CODIS loci plus two more
• Same 13 loci used by all labs • Tests measure the length of a piece of DNA
Basic terminology: Statisics • Random Match Probability: the probability that an randomly selected person from a given population of unrelated individuals would have the same profile as the evidence
• Combined Probability of Inclusion: The probability that a random person would be included in the observed mixture
Cold Hit • Why are cold hit cases statistically different from “probable cause” cases? Search through the databank gives the government thousands of opportunities to find a “match” The more profiles searched, the greater the chance of identifying an innocent person Databank has relatives and mixed racial groups
Database Searches and Cold Hits • Ten million tickets are sold for a lottery. What are your chances of winning if you buy: – One ticket? (1 in 10 million) – Ten tickets? (1 in 1 million) – 10,000 tickets? (1 in 1000)
• Among innocent people, 1 in 10 million has a DNA profile that would “match” the evidence. What are the chances the government will find a match if it compares the profiles of: – – – –
One innocent person? (1 in 10 million) Ten innocent people? (1 in 1 million) 10,000 innocent people? (1 in 1000) Five million innocent people? (1 in 2)
Cold Hit Statistics
• NRC I—test additional loci and report F for those loci only – Presumes ascertainment bias is a serious problem
• NRC II—report FxN, where N is the number of profiles in the database – E.g., if F=1 in 1 billion; N=1 million; then tell jury RMP=1 in 1000
• prosecutors and lab techs everywhere— ascertainment bias is not a problem, so just tell the jury F
Example of Database Match Probability NRC II—report FxN, where N is the number of profiles in the database
PEOPLE v. JOHN PUCKETT
RANDOM MATCH PROBABILITY:
1 in 1.1 million
DATABASE MATCH PROBABILITY: 1 IN 3
DATABASE MATCH PROBABILITY OR NRC II • THE KEY TO THIS STATISTIC IS THAT THE NUMBER GENERATED IS THE PROBABILITY OF FINDING A MATCH IN A DATABASE OF A GIVEN SIZE WHEN THE ACTUAL PERPETRATOR IS NOT PRESENT IN THE DATABASE
MIXTURES Subjective Interpretation
Mixture Interpretation • As the number of contributor’s grow, the data becomes harder to evaluate • Major versus minor contributors • Peak height ratios • Dropout • Dropin (LCN)
Mixture Interpretation ASSUMPTIONS • Number of contributors • Intimate samples – subtracting victim profile
“RESOLVED” OR “DEDUCED” PROFILE STATISTICS
Casework mixtures can be quite complicated • Example shows at least 4 donors • Other contributors unknown • Peak heights meaningless • Becomes difficult to exclude anyone
Mixtures
• More than two alleles at a locus may indicate a mixture • Number of contributors often unclear because of sharing alleles • Some labs rely on ‘peak height ratio’ • May be arbitrary: factors other than the quantity of DNA can effect peak height • Statistics used in mixture cases: may make debatable assumptions
Allelic Dropout D5S818
D13S817
D7S820
Evidence sample
Reference sample
• Peaks in evidence samples all very low – Mostly below 150 rfu
• At D13S817: – Evidence sample: 8, 8 – Reference sample: 8, 14
• 14 allele has dropped out • Tend to see with ‘marginal samples’
• Conclusion: Suspect included, But Why?
Allelic Dropout • Suspect Included Why? – – – –
We know allelic dropout occurs We can’t say it didn’t happen here (But can’t say it did) To be “conservative” we conclude that this is an inclusion
• Subjectivity – What is a peak, what’s not
• Examiner Bias – Included at all the other loci, therefore it must be him
A forensic lab assigned the major/minor alleles incorrectly @ THO1: major profile=6,6 and minor profile=8,9.3. Correct call is major profile=6,8 and minor profile=6,9.3 of 2-person mixture, 3:1 ratio.
Mixtures and Cold Hits
SAME SEARCH DIFFERENT RESULTS
EXAMINER BIAS AND MIXTURES SUBJECTIVITIY AND BIAS IN FORENSIC DNA MIXTURE INTERPRETATION, IN PRESS, SCIENCE AND JUSTICE ITIEL DROR AND GREG HAMPIKIAN
TASK: 17 ANALYSTS FROM SAME LAB EXAMINED THE SAME MIXTURE AND REFERENCE SAMPLES RESULT: 1 ANALYSTS INCLUDED SUSPECT 4 ANALYSTS CONCLUDED “INCONCLUSIVE” AND 12 ANALYSTS EXCLUDED THE SUSPECT
