Monthly Archives: July 2010

DNA Mythology

“The best model for the validation work is DNA, which was delivered full-blown to the forensic labs after years of validation in university laboratories.” Nature News of July 22, 2010, credits “Thomas Bohan, director of Portland, Maine-based consultants MTC Forensics and a past president of the AAFS [American Academy of Forensic Sciences]” with this capsule summary of the history of DNA evidence. But it is a capsule that no one should swallow — and that should stick in the craw of the Forensic Science Service, the FBI, and the commercial laboratories that spent years doing validation studies for DNA profiling. Dr. Bohan’s view is no more believable than the story of Athena bursting forth from Zeus’s forehead fully armed with weapons. Forensic DNA typing began in the laboratory of Sir Alec Jeffreys. What emerged was a normal infant that had a long way to go before attaining maturity. For the true story of the slow and bitterly contested process of validation of forensic DNA science, see Aronson (2007) or Kaye (2010).

References

Jay D. Aronson Genetic Witness: Science, Law, and Controversy in the Making of DNA Profiling (Rutgers Univ. Press 2007)

David H. Kaye, The Double Helix and the Law of Evidence (Harvard Univ. Press 2010)

Laura Spinney, Forensic Science Braces for Change, Nature News, July 22, 2010, doi:10.1038/news.2010.369

“Familial Searching”: Ten Questions and Answers

“Familial searching” of DNA databases is back in the news–big time. A California serial killer shot and strangled at least 11 victims, mostly in the 1980s. Thirteen years elapsed between two of the murders, but when the killings picked up again, a newspaper called him the “Grim Sleeper.” Last week, to great fanfare, Los Angeles police arrested a man whose DNA matches several samples associated with the killings. They found him as a result of a “familial search” of the state DNA database. Here, I consider ten questions that one might have about the investigative practice. (For earlier remarks on the subject, see “New York to Use Partial DNA Database Matches that Implicate Relatives,” Jan. 25, 2010, on this blog.)

1. What is “familial searching”?

Forensic DNA databases contain “STR profiles”–a set of numbers that represent features of DNA that tend to vary from one person to another. The profile comes from thirteen locations (“loci”) in the genome. The locations are not genes, and the DNA variations (“alleles”) there are associated with one’s ancestry. Some alleles are more or less common among some ancestral groups than others, but the forensic STRs are not all that good for inferring ancestry. About all that they are really good for is identifying individuals and checking into relatedness (as in parentage testing). Unrelated people have some alleles in common, but not many. At the other extreme, identical twins, being clones, have all their alleles in common. Other siblings have about a quarter or more in common. A parent and a child have about half or more in common. Therefore, when a DNA sample from a crime scene is almost –but not quite–a match to a particular individual in the convicted-offender database, the crime-scene could well have come from a full sibling, a parent, or a child. As one moves farther out on the family tree, it quickly becomes difficult to distinguish relatives from unrelated individuals.

In an ordinary database trawl, the crime-scene profile is checked against the many profiles from known offenders. If the numbers that constitute the two profiles match, then the past offender becomes the target of the investigation. Used in this fashion in the Grim Sleeper case, the DNA database trawl drew a blank. Nobody matched. [1]

Then police tried checking for near matches on the theory that possibly the Grim Sleeper had a brother or other close relative in the database. Figuring out which partial matches are good indications of relatedness, however, is not easy. The list of near matches usually includes far more nonrelatives than family members. Although the probability that a nonrelative will nearly match is much smaller than that for the near match to a single very close relative, there are far more unrelated people in the database. By chance, some of them will be near matches.) Further investigation must be done to cut the list down to a reasonable size. I shall call the process of generating a candidate list (of mostly unrelated people) “near-miss” searching.

2. When was it first used?

The first successful “familial search” occurred in 2004, in England. It involved not only near-miss searching, but low-template DNA analysis. (See “Disagreement on What to Call DNA Profiling with Really Small Samples Confuses Courts,” July 16, 2010.) The link to the defendant came from DNA on a brick. The defendant threw it off a footbridge on the M3 motorway. It went through a windshield, striking a truck driver on the chest. He managed to stop his truck before expiring from a heart attack. [2]

3. Does it give the police a list of family members to investigate?

No. It provides a list dominated by unrelated people. (See Question 1.) Some of them could have higher degrees of matching than an actual relative. In the Harmon case (Question 2), screening the search to white males, aged under 35, living in Surrey or Hampshire produced a list of 25 names. [2]

4. Is it “genetic surveillance”?

Not unless, like Alice said to Humpty Dumpty, “you can make words mean so many different things.” George Washington University Law Professor Jeffrey Rosen speaks of “permanent genetic surveillance.” [1] But generating a list of possible suspects is no more “surveillance” than is looking to see if a fingerprint from a crime-scene matches a fingerprint that is on file. Of course, the police could conduct on-going physical surveillance of individuals in the candidate list, and that is what they did in the Grim Sleeper case. An undercover team followed the father of the man whose DNA profile in the database was a near match. When he discarded a slice of pizza, they retrieved it. The laboratory determined that DNA on the pizza matched the killer’s. [3]

Whether police should be allowed to collect shed DNA rather than obtain a search warrant for a suspect’s DNA is a nice question. Courts routinely uphold the practice, but whatever one thinks about its constitutionality, it has no inherent connection to”familial searching.”

The adjective “genetic” also is misleading. The forensic identification profile has no substantial social or medical significance. No one is studying the genes of family members for clues to whether they will develop cancer or commit future crimes. (See Question 1.) Only by stretching the traditional meaning of “genotype” can forensic DNA profiling even be called genotyping. No genes are being studied. In short, the phrase seems to be a scare tactic employed by opponents of near-miss searching.

5. Will it cause innocent people to be convicted?

It is hard to see how. As with any DNA database search, the defendant’s DNA must be shown to match the full crime-scene DNA profile. Of course, errors can occur in any system, and there can be cases in which someone whose DNA is found at a crime-scene is not the criminal. But the chance of a false conviction is no greater than with an ordinary trawl that produces an immediate “cold hit.” Furthermore, the technology can help free innocent prisoners, as it did for Darryl Hunt in 2005. [4]

6. Is it guilt (or loss of privacy) by association?

No, being on the list of near misses is not evidence of guilt. (See Question 5.) But becoming a suspect is not pleasant and can result in pressure to give the police a sample of DNA. UCLA Law Professor Jennifer Mnookin has written that “those people who just happen to be related to criminals have not given up their privacy rights as a consequence of their actions. To use a search technique that targets them simply because of who their relatives are is simply not fair.” [5] But it is not apparent that there is any fundamental “privacy right” to be free from becoming a target of an investigation because of one’s associations with individuals who come to the attention of the police. Suppose that I commit a crime all by myself but I have a nosy neighbor who shadowed me. He gets caught committing a totally unrelated crime, and he bargains for a lower sentence by offering to rat on me. Would we insist that “those people like me, who just happen to be living next to nosy criminals have not given up their privacy rights as a consequence of their actions. To use a search technique that targets them simply because of who their neighbors are is simply not fair.”?

7. Does it violate the Fourth Amendment protections against unreasonable searches?

According to a Los Angeles Times article, unnamed “legal scholars” think the California policy is an unreasonable search that might run afoul of the Fourth Amendment. The paper also quotes “Tania Simoncelli, science advisor to the American Civil Liberties Union,” as asserting that “The fact that my brother committed a crime doesn’t mean I should have to give up my privacy!” [6]

Serious legal scholarship on the question is limited and unsettled (unless we count op-ed pieces and blogs as scholarship). A student law review note suggests that a plausible Fourth Amendment argument could be made. [7] Another student wrote that “the Fourth Amendment does not prevent familial DNA searches.” [8] The only published law journal article from a law professor concludes that it is “clear is that the Fourth Amendment, as currently interpreted, provides no privacy bar to the implementation of familial DNA investigation regimes.” [9] An article in press questions that conclusion, but only by stretching existing Fourth Amendment doctrine. [10]

Is there a plausible basis for applying the Fourth Amendment to bar inferences from DNA profiles legitimately in the possession of the government? A workable principle that would shield me from DNA information extracted from my relatives without my permission? Consider the following case: You have an identical twin brother. He robs a bank, is locked away in prison, and his DNA profile is put in an offender database. This can happen even though his DNA was not evidence in the bank robbery case and had nothing to do with that crime.

While your brother is out of circulation, you break into a house. cutting your hand on the glass of a window that you shattered to gain entry. A tiny bloodstain with your DNA on it is analyzed. The profile is compared to those in the database. It matches the one that is file perfectly — your brother’s — because identical twins have the same DNA sequences. But the police know that your brother was in prison when the house was burgled. They scratch their heads until they realize that he might have an identical twin with identical DNA.

So the police investigate you and find plenty of other evidence against you. Now you are facing trial. You move to exclude evidence that your DNA matches that in the bloodstain on the ground that this discovery is the result of an unreasonable search, arguing that “the fact that my brother committed a crime doesn’t mean I should have to give up my privacy!” Not only that, you contend that the rest of the evidence must be suppressed because all of it is the fruit of this illegal search.

If convicted-offender databases that include bank robbers are constitutional (as they surely are), how can this search infringe the Fourth Amendment? It is too bad that you and your brother share the same DNA profile, but the police have not forced you to surrender your DNA, and you have no right to stop them from checking your brother’s DNA to see if he might be responsible. By checking him, they learn something about you. You might not like it, but let’s face it, this probably is not the first time that your brother got you into trouble.

A counter-argument, however, is that the example assumes that the police have acquired the sample legitimately. Courts have used a balancing test in determining that taking DNA samples from offenders databases is reasonable under the Fourth Amendment. This balancing has not counted the interests of relatives in being free from suspicion. If near-miss trawling poses a serious enough intrusion on the privacy of relatives, then the system of databanking that allows this practice is impermissible. Although I am skeptical that the impact on relatives rises to this level, for now I shall leave this as an open question.

8. Does it target minorities? Is it “racial profiling”?

According to a New York Times reporter, “Those who oppose the technique argue . . . that it serves, in essence, as a form of racial profiling because a higher proportion of inmates are members of minorities.” [1] This is an abuse of the term “racial profiling.” Looking for investigative leads among known offenders and their associates is not like deciding to issue speeding tickets to blacks but not whites or pulling over a black driver just because he is in a white neighborhood.

Although near-miss searching is not racial profiling, the technique does have a disparate impact on minorities. So does an ordinary database search. Minorities are concentrated among prison inmates relative to their proportion of the general population. They are at greater risk for arrests. Therefore, they are concentrated in the DNA databases as well. Relying on DNA databases for investigative purposes perpetuates (but does not magnify) the relative disparity. The perception of a system stacked against minorities is a legitimate concern.

[A politically incorrect note: In terms of racial justice and fairness, one has to ask whether percentages of the general population supply the relevant benchmark for determining what is disproportionate. If the proportion of criminals in some minority groups is higher than in majority groups, those minorities should be overrepresented in the known criminal population. There is nothing wrong with targeting criminals regardless of race even if the effort nets a larger proportion of one race than another. There is something wrong a society that unfairly disadvantages minorities (leading them to violate the criminal law in larger numbers), that criminalizes conduct in one group without criminalizing comparable conduct in another, or that enforces criminal law along racial lines. Many factors are at work to produce the disparities in the criminal justice system. But one cannot conclude that a concentration of minorities in a database is intrinsically unjust, and the comparative justice argument against existing DNA databases is more subtle than the pragmatic one presented in the preceding paragraph.]

9. Does it violate the Fourteenth Amendment’s guarantee of equal protection?

No. It would take a sea change in Equal Protection doctrine to find any such violation. Only intentional discrimination is unconstitutional. Policies applied evenhandedly that do not use a racial classification but that burden some racial or ethnic groups more than others do not offend the Equal Protection Clause. The Supreme Court made this clear in the 1970s. To the extent that a white first-time offender is at less risk for questioning or apprehension than a minority first-time offender, near-miss searching has a disparate impact. As explained above, this is a matter of concern, but it is not a constitutional flaw.

10. Is it illegal in the absence of a statute explicitly authorizing it?

The strongest legal argument against near-miss database searching (as presently conducted) is that the legislature has not approved it. To be sure, one can argue that the practice falls squarely within the broad wording of DNA database statutes. These laws typically permit samples to be used for “identification.” A literal reading of the statutes should permit near-miss searching.

But I am not a fan of literalism, and figuring out which construction best advances the purposes of the statutes or respects the intent of the legislative body is not easy. When the statutes were written, legislators did not contemplate using the database samples and profiles to identify suspects outside of the database. It is one thing to vote for a bill that requires convicted offenders to submit DNA samples that could be used against them if they commit later crimes (or have committed earlier ones). It is another thing to vote for the bill knowing that the samples sometimes will be used to produce leads to family members who are subject to compulsory DNA sampling.The latter practice does not offend fundamental privacy rights (see Question 7), but in a society that prizes minimal government scrutiny and individualism, a legislator could consider it problematic.

At a minimum, transparency in the policies governing “familial search” is important. The detailed policy could come from the legislature in the first instance, or it could come from the executive branch and be subject to legislative disapproval or modification. In the absence of preempting legislation, police have an inherent power to use any constitutional investigative technique, including DNA database trawling. But the practices should be open for all to see.

References

1. Jennifer Steinhauer, ‘Grim Sleeper’ Arrest Fans Debate on DNA Use, N.Y. Times, July 9, 2010, at A14

2. Forensic Science Service, Craig Harman – Family DNA Link Offers Crime Breakthrough, 2008, at http://www.forensic.gov.uk/html/media/case-studies/f-39.html

3. Maura Dolan, Joel Rubin & Mitchell Landsberg, DNA Leads to Arrest in Grim Sleeper Killings, Los Angeles Times, July 8, 2010

4. Richard Willing, Suspects Get Snared by a Relative’s DNA, USA Today, June 7, 2005.

5. Jennifer Mnookin, Devil in the DNA Database, Los Angeles Times, Apr. 5, 2007, http://articles.latimes.com/2007/apr/05/opinion/oe-mnookin5

6. Maura Dolan & Jason Felch, California Takes Lead on DNA Crime-fighting Technique: The State Will Search its Database for Relatives of Unidentified Suspects in Hopes of Developing Leads, Los Angeles Times, April 26, 2008

7. Daniel J. Grimm, Note, The Demographics of Genetic Surveillance: Familial DNA Testing and the Hispanic Community, 107 Colum. L. Rev. 1164 (2007)

8. Jessica D. Gabel, Probable Cause from Probable Bonds: A Genetic Tattle Tale Based on Familial DNA, 21 Hastings Women’s L.J. 3, 26-27 (2010)

9. Jules Epstein, “Genetic Surveillance”– The Bogeyman Response to Familial DNA Investigations, 2009 U. Ill. J.L. Tech. & Pol’y 141, 165 (2009)

10. Erin Murphy, Relative Doubt: Familial Searches of DNA Databases, Mich. L. Rev. (forthcoming).

Disagreement on What to Call DNA Profiling with Really Small Samples Confuses Courts

In most U.S. jurisdictions, expert evidence derived from a scientific method that is not generally accepted (the “Frye standard”) or adequately validated (the “Daubert standard”) is inadmissible. Applying these standards properly requires distinguishing between a methodology and its application on a particular occasion.1/ An example of confusion over the methodology-conclusion distinction in forensic DNA testing involves what has been called Low Template (LT), Low Copy Number (LCN), or just “touch” DNA.

Unfortunately, scientists do not seem to agree on which term applies to what,2/ muddying the waters for the courts now being asked to rule on the admissibility of LT DNA profiling. The LT DNA profiling procedure amplifies DNA fragments from the standard STR loci, but the samples of DNA are really tiny. In criminal cases, they come from minute deposits of skin cells transferred by handling a gun, gripping a steering wheel, turning a doorknob, throwing a brick, or the like. With suitable techniques for extraction and amplification, STR profiles can be coaxed from as few as 5 to 20 cells. The most obvious strategy is to increase the number of PCR amplification cycles. The danger is that chance effects might result in one STR fragment being amplified much more than another. Parts of the full profile then could drop out, small peaks from unusual fragments at other loci might “drop in,” and a bit of extraneous DNA could contribute to the profile. Other protocols have been proposed for typing such small quantities of DNA.3/

In People v. Megnath, 898 N.Y.S.2d 408 (N.Y. Sup. Ct. 2010), the first published opinion on the need for strict scrutiny of “LCN DNA analysis,” id. at 410, a New York trial court wrote that because

the LCN DNA method of testing as performed by the OCME [Office of the Chief Medical Examiner] is basically the same technique as HCN [high copy number] DNA testing, with the exception of its increased amplification cycles, the Court finds that LCN DNA testing as performed by the OCME is not a novel scientific technique for the purposes of the Frye inquiry. Indeed, both the LCN and HCN forms of DNA testing require the same steps to be taken. These steps, namely extraction, quantitation, amplification, and electrophoresis are virtually identical in both HCN and LCN DNA testing.

Id. at 415. On this basis, this court concluded that rigorous review of the state of the science was unnecessary.

However, inquiring into the extent to which the variations on the usual extraction and amplification procedures represent the same method or a different one misses a more fundamental point. PCR-based STR profiling with capillary electrophoreses is generally accepted for the purpose of producing identifying profiles because experiments have demonstrated its validity and it fits into well established theories of chemistry and biology. It satisfies the validity standard of Daubert for the same reasons. But this foundation might not extend to the domain of the smaller samples. A light microscope works wells for studying bacteria but its magnification is not adequate for viewing much smaller viruses. For that purpose, an electron microscope is required. Radar can track airplanes or flocks of birds, but the signal-to-noise ratio is too low for it to be useful in tracking the flight path of a solitary, high flying butterfly. The radar equipment is identical, and operator is no less skilled at interpreting what he sees on the screen, but the procedure has not been validated (and would not be valid) for butterfly tracking.

Likewise, in the case of touch DNA, the relevant question is not whether the instrumentation and chemicals are identical or whether the analysts are using the same standards for interpretation. It is whether the system has been validated in the range in which it is being used. This is not a question about how well a validated or generally procedure worked on a particular occasion–an issue to be resolved on a case-by-case basis. It is a broadly applicable question about the ability, under the best of conditions, of the equipment and its operators to pick out a weak but true signal from the noise. Until this trans-case question is resolved, admissibility is unjustified under Frye and Daubert.

Fortunately, in Megnath itself, the court conducted a pretrial Frye hearing and found the modified procedure to be generally accepted as valid for touch DNA. Id. But in United States v. Davis, 602 F. Supp.2d 658 (D.Md. 2009), the government, responding to defendant’s effort to dismiss profiling of low template DNA as “the latest fad in DNA testing,” id. at 667, fought to avoid a Daubert hearing by claiming that the laboratory was using the same equipment and interpretive methods that previously had been held admissible as applied to large samples of DNA. The government expert maintained that if no additional cycles were added to the PCR step (and if no other modifications were made in the usual steps), then no LCN testing was occurring. Id. at 669.4/ The defense expert, Dan Krane, responded by defining “LCN testing” so as to depend on the state of mind of the laboratory technician5/ or on the quantity of DNA at the point of PCR amplification.6/ The district court avoided “making a finding with regard to the dueling definitions of LCN testing advocated by the parties” by finding that “the amount of DNA present in the evidentiary samples tested in this case” was in the normal range.7/

This semantic battle over “the proper, scientifically accepted definition of low copy number testing,” Davis, 602 F. Supp.2d at 667, is pointless. It makes no difference whether STR profiling of very small samples is denominated “LCN testing,” “HCN testing,” “STR profiling,” or any other concatenation of letters. The issue is how well the procedure, whatever its called, works with samples that are so small that the stochastic effects in PCR amplification could produce a false profile. There is no single sample size at which excessive amplification of the wrong DNA fragments and the absence of amplification of the right fragments occur. Ever since PCR-based systems were introduced, critics of DNA testing have argued that such DNA typing is subject to these problems. See David H. Kaye, The Double Helix and the Law of Evidence 182-83 (2010). It is not a reason to exclude all PCR-based analyses, but the proponent of the evidence must be able to show that in the relevant sample size range, it is possible to ascertain the signal reliably. This is the trans-case, scientific question of methodology to be answered under Frye or Daubert.

Notes

These remarks are adapted from a discussion in The New Wigmore, A Treatise on Evidence: Expert Evidence (David H. Kaye, David E. Bernstein & Jennifer L. Mnookin eds., 2d ed. 2011) (in press).

1. See The New Wigmore, A Treatise on Evidence: Expert Evidence ch. 7 (David H. Kaye, David E. Bernstein & Jennifer L. Mnookin eds., 2004).

2. Whether the terms are truly synonymous is itself the subject of contention. See, e.g., Bode Technology, Touch DNA–Overview, http://www.bodetech.com/technologies/touch-dna/touch-dna-overviewd, last visited June 30, 2010 (insisting that “Touch DNA is not Low Copy Number (LCN) DNA.”) (emphasis in original).

3. See, e.g., John Buckleton & Peter Gill, Low Copy Number. in Forensic DNA Evidence Interpretation 275 (John S. Buckleton et al. eds., 2005); Pamela J. Smith & Jack Ballantyne. Simplified Low-Copy-Number DNA Analysis by Post-PCR Purification, 52 J. Forensic Sci. 820 (2007).

4. But in the unreported case of United States v. Williams, No. CR 05-920-RSWL, 2009 WL 1704986 (C.D. Cal. June 17, 2009), a different prosecution expert “declared that, without any increase in the amplification cycles, the amount of input DNA is low enough to constitute LCN when the amount is less than .1 ng or .2 ng of input DNA.” Id. at *2.

5. 602 F. Supp.2d at 669 (“The one feature that all LCN processes have in common is the knowledge or expectation that less than the recommended amount of template DNA is being used.”).

6. Id. (“Using small amounts of template DNA (even without making any changes to the testing process itself) is all it takes for something to be in the LCN category.”).

7. Id. See also Williams, No. CR 05-920-RSWL (agonizing over definitions but concluding that the sample quantity was not low enough to constitute “LCN” under either party’s definition).