Scientists now have the ability to identify an indefinite number of physical traits including height, eye color, sex and race from a trace of DNA material. Recent breakthroughs in the Human Genome Project (HGP) mandate an expansion of DNA evidence as an investigative tool.
A few years from now we’re going to have figured out so many traits that a criminal might as well leave his driver’s license at the scene of the crime.
An unidentified girl, presumed to be 3 or 4 years old, was found decapitated in Kansas City, Missouri, in 2001. Local residents had given her the name “Precious Doe” in order to humanize her tragic fate and draw public interest to her case. The police sent her DNA to be analyzed for clues to her ethnic heritage. Results of her DNA test indicated that Precious Doe was of mixed ancestry, approximately 40 percent Caucasian and 60 percent African American. From the tests, forensic specialists in DNA analysis estimated that Precious Doe had a white grandparent. Police also had a tip from an Oklahoma man who said that he was a relative of the slain girl and knew who killed her. They also sought out people who had failed to report the disappearance of a child. From the tips they narrowed the suspects to one woman who had not reported her child as missing and had one black and one white parent. When they ran a DNA test of the woman, they discovered that she was the biological mother of Precious Doe. Both she and her husband, the girl’s stepfather, were charged with her homicide. Writing about the case for USA Today, Richard Willing noted, “Advances in DNA testing are allowing investigators to learn more about suspects whose profiles are not in the databases. Tests that can identify a suspect’s ancestry are being used not to identify the suspect by name, but rather to give police the idea of what he or she looks like.”3
The term “DNA profiling” first came into use when Alec Jeffreys discovered how to use DNA samples to determine identity or paternity. “DNA profiling” became synonymous with terms like “DNA typing,” “DNA fingerprinting,” “DNA identification,” and “forensic DNA testing.” In this sense DNA profiling means sequencing the DNA loci that can be used to compare two samples to determine if they came from the same person. The DNA profile is the set of 26 numbers that characterize the short tandem repeats (STRs) at 13 loci in the human genome (see chapter 1).
However, the genetic age has brought with it a new usage of DNA profiling that is similar to visual profiling of crime suspects. In a visual profile an eyewitness identifies some characteristics of the alleged perpetrator of a crime, such as height, hair color, distinguishing marks, skin color, and body size and shape. A composite sketch (profile) is created from bits of eyewitness information. In forensic DNA phenotyping forensic scientists analyze DNA from a crime scene in an attempt to determine certain aspects of the individual’s physical characteristics. This “second generation” of DNA profiling goes well beyond examining the 13 loci that are used and stored in the U.S. Combined DNA Index System (CODIS) database. Instead, one or more genetic tests are run to glean information from the DNA that might provide clues about the suspect’s physical features. Lindsy Elkins describes DNA phenotyping as follows: “Scientists can now discern from DNA a virtually indefinite number of physical traits possessed by an individual from height, eye color, sex and race, down to the shapes of a person’s toes. In addition, genetic typing permits inferences as to inherited disorders and may offer clues to facial or other bodily features.”4
Forensic DNA phenotyping is not without socioethical concerns. These include the following: Does DNA phenotyping by police infringe on civil liberties? Will DNA phenotyping exacerbate racial or ethnic stereotypes? Will it help law enforcement solve crimes? Will DNA phenotyping reduce the use of prejudicial and subjective (based on eyewitness reports) profiling?5 How scientifically credible is phenotyping by DNA? Because the technology of DNA phenotyping is still in its early stages of development, the answers to these questions must be tentative.
Linking Genotype to Phenotype
Lindsy Elkins in her law review article on physical profiling based on genetics is far too optimistic about the relationship between the DNA and the phenotype (“down to the shapes of a person’s toes”).6 There are significant uncertainties associated with predicting racial or ethnic origins or physical qualities on the basis of DNA. The remaining sections of this chapter discuss the current possibilities and limitations for “phenotyping” DNA. Can we draw inferences about physical appearance or the medical condition of an individual from his or her DNA? What are the possibilities of creating a physical profile of an individual exclusively from crime-scene data?
Three research programs in genetics have provided the scientific foundations behind forensic genotype-to-phenotype profiling: ancestral testing, behavioral genetics, and medical genetics. In each case scientists seek to find links between components of the human genome and an individual’s physical characteristics, behavior, or response to environmental factors such as medication or antigens.
Ancestral Genotyping
The Human Genome Diversity Project, an outcome of the Human Genome initiative, has been used to identify genetic patterns that reveal one’s biogeographical ancestry. Population-specific alleles (PSAs) enable geneticists to distinguish the genotypes of a select number of ethnic populations, such as Europeans, Africans, Native Americans, and Asians. And although there are more genetic variations within ethnic/racial groups than between groups, some researchers have found that there are sufficiently stable genetic segments (haplotypes) that can distinguish among several ancestral population groups. These are called ethnic geographical markers of ancestral origin, or Ancestry Informative Markers:
Ancestry informative markers (AIMs) . . . are autosomal genetic markers that show substantial differences in allele frequency across population groups. . . . Factors such as isolation by distance, range expansions, land bridges, maritime technologies, ice ages, and cultural and linguistic barriers have all affected human migration and mating patterns in the past and have therefore shaped the present worldwide distribution of genetic variation.7
Current ancestry tests examine upwards of 176 genetic markers in which the DNA varies at only one position. These single-nucleotide polymorphisms (SNPs) have been found to occur more frequently in certain population groups than in others—the result of centuries of geographical separation and group intermarriage.8
Because most people have mixed biogeographical ancestral origins, the programs that yield information of this kind provide percentages (such as 80 percent African and 20 percent European). The composite percentages typically cannot yield definitive information about the phenotype of specific individuals falling within these biogeographical categories, but the question of probabilities for populations is left open.
According to Susanne Haga, assistant research professor at the Duke Institute for Genome Science and Policy, because admixture confounds the prediction of ancestral origins, “a substantial gap remains between ancestry and/or race and physical appearance. . . . Knowing an individual’s race [percentage of racial origin] can be misleading if used to predict certain physical traits.”9 When someone is told they are 60 percent Asian and 40 percent European it is based on large segments of DNA. We do not know what the 60 percent Asian DNA will express itself as; therefore we cannot from those percentages say anything about the phenotype of the individual. But if we had probability figures, such as “90 percent of the people who are tested by DNA ancestry as 60 percent Asian have ‘yellow’ skin tone,” then we may be able to draw some probabilistic predictions about phenotype.
Ancestral Genomics in Law Enforcement
Ancestral genomics is beginning to find a place in law enforcement. In March 2003 a company known as DNAPrint Genomics of Sarasota, Florida, analyzed the DNA of a serial killer in Louisiana using a genetic ancestry technique trademarked as DNA Witness. The company concluded that the killer’s “biogeographical ancestry” was 85 percent sub-Saharan African and 15 percent Native American. Until then, police had been seeking a Caucasian male in a pickup truck. The analysis of biological evidence at the crime scene by DNA Witness concluded that the Louisiana task force’s search was misguided, and that the individual they were looking for was more likely to be a “lighter skinned black man.” This description was inferred from probabilistic ancestry percentages revealed in the perpetrator’s DNA. The Louisiana police were dubious about the reliability of DNA Witness in profiling the serial killer. They sent the company 20 other DNA samples of individuals of whom they alone knew the ethnic or racial identification. According to the research director at DNAPrint Genomics, the company correctly identified the ancestry of all the samples.10 As a result of the analysis, Louisiana police shifted the focus of their investigation and identified Derrick Todd Lee, an African American man, as a possible suspect in the crime. Lee became the first person in the United States to be identified as a possible murder suspect through the use of a DNA test that racially profiled his DNA. On August 11, 2004, Lee was convicted in the first of a series of murder and rape cases.
DNAPrint Genomics used its success in Louisiana to aggressively market its services to police departments, investigators, and agencies.11 Toward the end of 2004 it started offering “RETINOME,” a genetic test to infer eye color, to law-enforcement agencies in addition to ancestry testing.12 Here is how the company described its forensic profiling services:
Testing DNA to create a physical description from crime scene DNA and providing a photo database array of representatives closely matching the analyzed DNA, allows detectives a means of describing “persons of interest.” This presumptive test method is a new market based on evolving DNAPrint™ Genomics technologies. Common hereditary traits such as skin pigmentation, eye color, hair color, facial geometry and even height can be predicted through analysis of DNA sequences. This can be done indirectly, through an extensive knowledge of ancestry admixture or for certain genetic traits, directly through knowledge of the underlying genes. At DNA-Print™ we use both methods. Our goal is to continue to lead the field of forensic presumptive testing using our DNA Witness™ line of products and services.13
The DNA Witness software technology was contracted out by the Boulder, Colorado, Police Department in 2004 to develop suspects in the highly publicized investigation of the rape-murder of a 23-year-old woman. In December 1997 Susannah Chase was walking home after getting a pizza when she was savagely raped and beaten to death with a baseball bat. When her murder investigation failed to lead to any suspects, it was put in the cold-case file. Six years later, in December 2003, the Boulder Police Department contacted DNAPrint Genomics to acquire some phenotypic information from the male sperm preserved from the victim’s body. In its January 2004 report the company stated that the source of the DNA was a person of Hispanic or Native American background. According to Boulder police chief Mark Beckner, “This technology gave our detectives a focus and direction that turned out to be right on the mark.”14
Police arrested a prime suspect in January 2008 by the name of Diego Olmos-Alcalde, a Chilean native who had an uncertain U.S. immigration status. DNAPrint Genomics appeared to have correctly identified the Hispanic ancestry of the person whose sperm was found at the crime scene. But what role did the ancestry information have in solving the crime? Did the phenotype “Hispanic or Native American” help police narrow the lead to the prime suspect?
As it turned out, it was not the phenotyping of the DNA that solved the case, but a match on CODIS, albeit a delayed one. The suspect was known to both Colorado and Wyoming police authorities for other felony arrests. He was arrested in 1998 in Denver on charges of attempted sexual assault and carrying a concealed weapon. However, prosecutors dropped the sexual assault charge in exchange for a guilty plea to the weapon charge.15 In May 2001, while in Wyoming, Olmos-Alcalde was given a 12- to 20-year prison term for kidnapping a Cheyenne woman in a parking lot near her apartment. Because of errors made by the trial judge, the Wyoming Supreme Court overturned the conviction and ordered a new trial, whereupon Olmos-Alcalde was resentenced in September 2004 to 7 to 10 years, with credit for time served.16 The defendant was paroled in the summer of 2007.
When the Denver police got around to loading the crime-scene DNA profile into CODIS, there was no match, that is, not until six months after Olmos-Alcalde was released from prison. If Olmos-Alcalde’s DNA profile had been uploaded to CODIS as soon as he entered prison, the police would not have needed the services of DNAPrint Genomics. There are two likely explanations for the delay. It is possible that his DNA was not profiled when he entered the Wyoming penal system but only after he was released. Alternatively, the profile of his DNA was not loaded into CODIS until he got out of prison or after some combination of delays. One report stated that officials were not sure when his DNA was profiled: “Melinda Brazalle, spokesperson for the Wyoming Department of Corrections, said the Department’s policy calls for collecting DNA from inmates during the intake process. But she wasn’t certain when Alcalde’s genetic information was obtained.”17 Colorado loaded the DNA sample taken from Chase’s body into CODIS as early as 2002 without a match, while Wyoming entered Olmos-Alcalde’s profile into CODIS in January 2008.18 The reason for the delay was the backlog of samples awaiting DNA processing. At the time there was a backlog of about 180,000 federal convicted-offender samples awaiting DNA processing and about 50,000 samples that were already processed but were waiting to be entered into CODIS.
There are two points to this story. First, DNAPrint Genomics’s ancestry analysis did not help solve this case; rather, it was solved from a DNA profile match in CODIS. Second, had there been no backlog in processing convicted felon DNA, police would have had no reason to consult DNAPrint Genomics.19
When the National DNA Advisory Board was considering the alleles it would recommend for use in forensic testing of crime-scene samples, its members made a deliberate choice to avoid any sequence that had known phenotypic properties or that disclosed ancestral origins. According to Ranajit Chakraborty, a board member and director of the University of Cincinnati’s Center for Genome Information, “In 1997, when members of the national DNA Advisory Board officially selected the gene markers for DNA evidence matching, they could have included a few markers associated with ancestral geographical origins (European, East Asian, Sub-Saharan African), which are a good indicator of race and ethnicity.” But the board instead decided against using racial markers because of the political sensitivity they represented.20 During the period in which CODIS was being developed, there was a heightened sensitivity both in law enforcement and among scientists that there were privacy issues involved in decoding people’s DNA that law enforcement should steer clear of. Although it was the board’s noble intention to use DNA exclusively for identification purposes, scientists continue to investigate methods that link allele variations in the STRs with racial and ethnic ancestry.21
Behavioral Genetics and Profiling
Research in behavioral genetics seeks to find links between genotype and certain human behaviors, specifically, but not exclusively, criminal behaviors.22 Thus, if genes could be strongly correlated with a person’s “fits of anger” or one’s pedophiliac tendencies, police could use genetic screening to narrow—or generate—a list of suspects in the search for a perpetrator of a violent and/or sexually deviant crime. Claims made by behavioral geneticists have sparked vigorous debates. Some scientists have critiqued the underlying science, pointing out the inherent limitations in studies of simple correlations that occur between genetic factors and complex behaviors.23 Others have cautioned that the miscommunications and misapplication of behavioral genetic research to policy could result in grave social consequences, especially in the area of criminal justice.24 Still others view the new field of behavioral genetics as a reincarnation of widely disavowed beliefs in genetic determinism and eugenics.25
In 1965 a study published in Nature found that a significantly higher number of inmates in a prison hospital in Edinburgh, Scotland, described as “dangerously violent” had an extra Y chromosome (XYY males), compared with the general population.26 The authors hypothesized that the presence of an extra Y chromosome produces extra aggressiveness and concluded that this condition increased the chances that an individual would be institutionalized. In 1970 President Richard Nixon’s personal medical adviser suggested that the country make use of this science, proposing a massive program of genetic screening for every 6-year-old to detect the “criminal potential” of preadolescents. Moreover, he suggested that every “hard-core 6-year-old” be sent to “therapeutic” camps where they could learn to be “good social animals.”27
The XYY study and other similar early studies lacked a satisfactory control group. When the studies were replicated with the proper methodology, scientists learned that XYY males tended to be taller, less intelligent, and more hyperactive but not necessarily more violent than their XY counterparts.28 The search for biological markers for social deviance has been a central theme in behavioral genetics. If such markers were discovered and their reliability were demonstrated, their forensic use would be nearly impossible to prevent so long as the entire human genome remains within the reach of law enforcement.
Another genetic marker called MAOA deficiency has been linked to violent behavior. A 2002 study published in Science by Avshalom Caspi and his colleagues looked at the DNA of 1,037 children who had participated in a 23-year survey of health and development conducted in New Zealand. Caspi and colleagues observed that a particular polymorphism—a gene on the X chromosome that regulates the production of the enzyme monoamine oxidase A (MAOA)—tended to moderate the effect of childhood maltreatment. That is, children with a version of the gene that caused greater production of MAOA were less likely to respond to maltreatment by lashing out and developing antisocial tendencies, while children with the low-MAOA-producing version of the gene had a higher rate of violent behavior. The MAOA gene oxidizes a chemical tryptophan into serotonin, one of several neurotransmitters (with dopamine and norepinephrine) that it uses to send signals. Mutations in MAOA can alter the amount of serotonin produced in the brain and biochemically affect emotion and behavior. Caspi and colleagues’ study makes no broad claims about genetic determinism but rather asserts that genetics may lend a partial explanation for differing behavioral responses to childhood maltreatment and concludes with the conjecture that “these findings could inform the development of future pharmacological treatments.”29
The MAOA gene could play several roles in criminal justice. Defense law-yers may seek to use MAOA testing results to support their client’s claim of mitigating circumstances in a violent crime (i.e., genetic defense against culpability). Alternatively, detection of an MAOA mutation in a crime-scene sample can lead police on a search for individuals with the defect, culling medical information for the match or using the result to profile individuals who may have a history of aggressive behavior. MAOA screening could become a factor in sentencing determinations or a requirement for consideration for early parole. Correctional facilities might prescribe routine MAOA testing for prisoners and forcibly administer medication to those presumed to be prone to violence on the basis of that testing. A 2002 Consensus Report by the Council of State Governments found that “staff at many correctional facilities have overrelied on the use of psychotropic medications and, in many cases, sedative-hypnotic medications, simply to pacify and to control inmates with mental illness and others believed to be disruptive.”30 Medication is often administered without adequate psychiatric evaluation and follow-up counseling and is often the sole treatment for mental illness and even nonclinical behavioral problems.31
The potential applications of behavioral genetic research to the criminal justice system raise profound questions for individual liberty and social justice. Past attempts to apply behavioral research to social policy—as evidenced most poignantly by the eugenics policies of the early to middle parts of the twentieth century—have been dismissive of the rights of individuals and vulnerable groups. Despite the increasing precision of molecular biology and repeated calls for caution about making social claims based on such research, there is a recurrent tendency to “biologize” crime and antisocial behavior and to use behavioral science as a justification for inequality and the promotion of new measures of social control. Therefore, we should be wary of any applications of behavioral genetics to the criminal justice system, whether for the narrow purpose of testing crime-scene samples on a case-by-case basis to predict behavioral traits of individual suspects or for the far broader purpose of characterizing the future dangerousness of the prison population or the general citizenry.
Medical Phenotyping
Medical geneticists seek to identify genes or multiple genetic loci (genetic markers) that correlate with individual disease states, predisposition to disease, or drug sensitivity. According to Wojciech and colleagues, “Studies developing genotype to phenotype correlations have advanced rapidly in medicine.”32 According to the National Institutes of Health, genetic tests are currently available for more than 1,700 diseases.33 In 2004 Hui Huang and colleagues utilized nearly 1,200 human disease sequences in a study.34 Some of these genetic mutations are asymptomatic, such as the Tay-Sachs trait, where only one copy of the mutation will not trigger disease manifestations. Others, such as Duchenne muscular dystrophy, which afflicts boys whose mothers are carriers, are severely disabling and prevent a person from living an independent life. Between these extremes are many cases where the genetic mutations may be connected to a disease of late adult onset where there are different intensities of disease expression. If DNA at a crime scene were analyzed for some or all known disease-related genetic mutations and there were a positive result for one that required medical treatment, police could track down potential suspects from hospital or pharmacy records of individuals treated for the relatively rare condition.
Consider the following scenario:
A series of burglaries take place in a small town in Virginia. DNA evidence in the form of a blood sample is collected from a window ledge of one of the homes. The Virginia authorities run the DNA against the state database; no match is found. They then send the sample to an outside laboratory for further testing. The lab runs a series of genetic screens on the sample and finds that the sample contains all four of the mutations that are among those most commonly associated with Gaucher’s disease. Upon receiving this analysis, Virginia authorities contact the National Gaucher Foundation and learn that most people living with Gaucher’s disease receive biweekly enzyme-replacement treatments. Furthermore, these treatments are administered at only two locations in Virginia, the Children’s Hospital and the University of Virginia. They contact the treatment centers and request a list of all individuals who have been treated for Gaucher’s disease over the last five years. The Children’s Hospital complies and turns over the records, but the University of Virginia refuses, claiming that this is private medical information that cannot be released.
To our knowledge, this scenario has not occurred. However, the facts about Gaucher’s disease and the locations of the treatment centers are accurate. In their pursuit of a felon, can and should the police be permitted to explore medical information from the genome of crime-scene evidence? Currently there is nothing to stop law enforcement from using whatever means it chooses to identify and analyze evidence obtained from the scene of a crime, including blood, tissue, hairs, and semen—that is, any biological materials. Crime-scene evidence does not possess any rights or privacy. The person who left that evidence at the scene or who is the source of the biological materials found at the scene has legal rights and privacy rights under the Fourth Amendment, but those rights do not include a protection against police acquiring information from the shed DNA. Therefore, criminal investigators can use whatever techniques they choose to find suspects from the evidence left at the crime scene. Some law-enforcement agencies have even resorted to hiring psychics who claim to have extrasensory powers that can provide information from objects found at the crime scene.35
If investigators are allowed to obtain scientific evidence in the form of mutations that are strongly correlated with Gaucher’s disease from discarded DNA at a crime scene, they then have a phenotypic clue to narrow the field of possible suspects. Can police gain access to medical records from hospitals, physicians, or medical centers for an undesignated and unspecified number of individuals who meet a specific medical criterion? Should this be considered a normal part of police investigation, or is this an intrusion on medical privacy that must be accompanied by a court warrant? Is this a type of “medical dragnet” without informed consent?
When police suspect that a fleeing suspect has been shot, they contact hospitals and seek information on whether someone was recently treated for gunshot wounds. Why would contacting local hospitals for information about people who are treated for Gaucher’s disease be any different? One answer is that when a person has a gunshot wound, he cannot be expected to have an expectation of privacy because he has entered the hospital in full view of everyone in his direct sight and is expecting emergency attention, perhaps to save his life from loss of blood or from damage to an organ. The situation is different for his medical information that may reside in hospital records.
Federal law does protect individual privacy in regard to personal medical records, but those protections are not absolute. If there is probable cause that an individual may have committed a crime, police can usually obtain a court warrant to gain access to medical records of an individual suspect. Confidentiality of medical records is established under the Health Insurance Portability and Accountability Act (HIPAA, passed on August 21, 1996). Applied to public health information, confidentiality means that information or data are not made available or disclosed to unauthorized persons and without proper cause. However, HIPAA contains a broad exception that allows disclosure of protected health information to law-enforcement officials, not only in compliance with a court order or grand-jury subpoena but also in response to an administrative subpoena, summons, or civil investigative demand—all legal instruments issued without judicial review.36 Broad administrative discretion is given to those with stewardship over health information at the hospitals in determining how to respond to written requests from law enforcement for patient records. HIPAA also allows health-care providers to disclose to law enforcement, on request, a broad array of identification information, including name, address, Social Security number, blood type, date of treatment, and a physical description.
When there is individualized suspicion and probable cause, police generally have no problem obtaining a court warrant for a suspect’s medical records. But in the case outlined, there is no individualized suspicion and therefore no probable cause against any single Gaucher’s patient. Should law enforcement be able to acquire information about all people of a certain geographical region who are treated for Gaucher’s disease because the crime-scene DNA has been medically screened and shown to have the alleles for the disease?
Judges do not typically give police investigators warrants when they are on a fishing expedition without probable cause that an individual or some small number of individuals is suspect. So our investigators may not be successful in getting a “wide-net” warrant from the courts to obtain the medical records of all Gaucher’s patients in an area. Should police be legally permitted to access medical records at Children’s Hospital and the University of Virginia without court warrants on the basis of the finding that DNA at the crime scene (which may or may not be the perpetrator’s) has the Gaucher mutation? Will the privacy of medical records keep the police from obtaining the information about males treated for Gaucher’s disease without a warrant? Will health centers turn over the information under current federal medical privacy statutes?
In our scenario the police are undertaking a kind of medical dragnet. Is a medical dragnet without informed consent ethically and/or legally justifiable? Should the same informed-consent principles hold for a medical dragnet that we expect to be in place for a DNA dragnet? Suppose that Children’s Hospital turns over to the police the names of the male patients who have been treated there for Gaucher’s disease over a period of 10 years. If police then narrow their suspects to three males on the basis of treatment for Gaucher’s disease and a suspected age range, is that reasonable suspicion to get a warrant for their DNA? These questions remain to be answered by the courts.
Predicting Sex, Hair Color, Eye Color, and Skin Color from DNA
Companies like DNAPrint Genomics that began providing products and services to the criminal justice sector for correlating phenotype with segments of DNA focused on a few genes and chromosomes where there had already been published studies that linked DNA alleles to physical characteristics. The most definitive trait prediction from DNA is sex. This can be done by chromosome analysis because the male has an X and a Y chromosome, while the female has two X chromosomes. The sex of a DNA donor can also be determined by a gene called the amelogenin sequence, which is present on both the X and Y chromosomes. But the genes on these chromosomes have different sizes, and those sizes can be read off a DNA analyzer.37 There are also male-specific genes (e.g., SRY) and female-specific genes (AR), which can be analyzed by PCR techniques.38 The amelogenin genetic analysis has been used in forensics and prenatal diagnosis.
According to Elkins, “Genetically-derived trait information may be superior to human-derived trait information because, unlike humans, machines cannot be fooled by changes in physical appearance.”39 This may be true, but the practical value of predicting appearance is less clear. For example, let us suppose that a witness described a possible suspect in a crime as having blond hair. Let us also imagine that the perpetrator of the crime left DNA at the crime scene, and a DNA test revealed that the suspect in fact had brown hair. Indeed, such a test might have greater reliability for natural hair color than an eyewitness report. But how helpful would this information be in tracking down the suspect if in fact the individual had dyed his hair?
In terms of inferring human nondisease physical characteristics from DNA left at a crime scene, we probably cannot do much better than the phenotype red hair/light skin. Red hair has been linked to variants of a single gene called MC1R, which encodes the melanocortin-1 receptor. Receptors reside at the surface or in the nucleus of cells and are acted on by specific proteins (such as hormones) to produce other proteins through the mechanism of DNA transcription. The action of a stimulating hormone called alphamelanocyte (aMSH) on the receptor MC1R controls the switch between a red/yellow substance (phaeomelanin) and the black/brown substance (eumelanin). People with red hair produce more phaeomelanin than brown- or black-haired people. One study found 12 variants of MC1R. Of those, 8 were associated with red hair. The authors found that for 96 percent of individuals they studied, those with 2 of the 8 red-hair-causing mutations had red hair. Two of their subjects who had the mutations but were not red haired (either blond or light brown) described themselves as having had red hair in their youth.40
Writing in the Journal of Forensic Sciences, Branicki Wojciech and colleagues noted that “determination of one of the [relevant] MC1R variants in the homozygous state or heterozygous combination can be considered a strong indicator that the sample donor has red or strawberry-blond hair and fair skin.”41 The relationship between certain genetic variants in MC1R has been reproduced and validated.42 From a forensic viewpoint, if a DNA sample shows two MC1R sequence mutations, it is a good bet that the donor of the sample has or had red hair. If the DNA shows that the individual is homozygous (has two copies) of the allele that is not associated with red hair, then it is a good bet that the donor of the DNA does not have red hair. The presence of one mutation would be inconclusive.43
Iris color is largely a genetic rather than an environmental trait. Studies of twins show high correlations (about 85 percent) of iris color between homozygote twins. How much pigment a person expresses in his or her iris is linked to three SNPs, also known as single-letter variations, in a DNA sequence near a gene known as OCA2. According to a 2007 study in the journal Human Genetics, OCA2 is the major human iris-color gene, and SNPs within this gene can accurately predict melanin content from DNA.44 Among the individuals carrying the same SNP sequence in all three locations on both copies of the gene, 62 percent were blue eyed.45 Because iris color does not change with age or sunlight, it is viewed as a stable and predictable trait that could be useful for crime-scene DNA profiling. However, with only slightly more than 50 percent genotype-to-phenotype predictability, phenotyping for iris color has limited reliability.
Skin color is considered a polygenic trait. This means that many genes in different loci of the human genome are responsible for pigmentation of the individual. There may be 30 to 40 genes responsible, which may act both additively and nonadditively in producing a spectrum of pigmentation colors that are found within the human family. Others have speculated that there are more than 120 genes that play some role in skin pigmentation. Variations of the color of human hair and skin are determined by the amount, density, and distribution of the two components making up the pigment melanin, which is produced in specialized cells known as melanocytes. The synthesis of two organic polymers, namely, eumelanin (dark brown/purple/black pigment) and phaeomelanin (yellow to reddish brown pigment), determines the amount of melanin in the skin. The exact role of the genes responsible for producing the ratio of these pigments (eumelanin and phaeomelanin) is not fully understood. Much of what is known about human pigmentation has been learned from mouse studies.46 Not only are the genetics and chemical pathways of melanin synthesis complex, but environmental factors in skin pigmentation also play a role.47
In his book Molecular Photofitting Tony Frudakis reports that an individual with each of nine major variants of a gene MC1R will more likely than not express a fair or pale skin phenotype relative to the average European. But he adds that “we need to be able to measure more than just the MC1R genotypes to make inferences and because skin color varies so much with biogeographical ancestry (on a continental level), the approach that could be useful is the admixture mapping approach.”48 Admixture mapping is a tool for uncovering genes that contribute to complex traits. It involves examining the gene frequencies of two or more genetically diverse intermating populations (admixture populations).
Given the complexity of the genetics and the gaps in our knowledge of human skin pigmentation, we cannot yet go directly from DNA to skin color. The indirect route to specifying skin color from genotype is through ancestry analysis, as previously discussed. However, ancestry analysis will yield mixtures of a person’s biogeographical heritage; skin pigmentation cannot be predicted determinatively from that knowledge.
Troy Duster reported that the forensic science laboratory in Birmingham, England, claimed that its DNA test can distinguish between “Caucasians” and “Afro-Caribbeans” in nearly 85 percent of cases.49 The original work cited for this result was published in 1993.50 Bert-Jaap Koops and Maurice Schellekens note that the differential power of phenotyping can result in racial bias:
The use of characteristics for a composite drawing or description of the perpetrator of a crime may lead to stigmatization of certain groups within society. . . . This is an even greater risk if one ethnic background can be better determined than others, as the FSS [Forensic Science Service, United Kingdom] indicates is the case with Afro-Caribbeans. This might lead to relatively more cases involving Afro-Caribbeans, simply because they can be determined where other DNA types would yield inconclusive information.51
This is the DNA counterpart to the old joke, “Why are you looking under the streetlight for your lost key?” The answer: “Because that’s where the light is.”
Tony Frudakis, formerly of DNAPrint Genomics, also supports the use of DNA for developing “facial geometry” of an individual. This might include facial bone structure as well as the size and shape of the face. In a project at University College London, scientists scanned the faces of hundreds of volunteers in order to find correlations between digitized facial geometry and genetic markers, but the project was abandoned in 2000 when it proved too complex. One German forensic biologist commented that we may never be able to fully reconstruct a suspect’s face from genes alone.52
Policies on Phenotyping DNA
The debate over forensic DNA phenotyping has been more acute in the European Union than it has in the United States. Only three states, namely, Indiana, Rhode Island, and Wyoming, disallow by statute the use of DNA submitted to their data banks for purposes of obtaining phenotypic information about the DNA sources.53 For example, the Wyoming law states that “information contained in the state DNA database shall not be collected or stored for the purpose of obtaining information about physical characteristics, traits or predispositions for disease.”54 But this does not restrict phenotyping crime-scene DNA to generate information about suspects before the DNA enters the data bank. Michelle Hibbert surmises that the vast majority of states permit DNA trawling in order to allow police investigators to trace unknown suspects or anonymous murder victims.55
At the federal level, as discussed earlier, HIPAA provides individuals privacy protection in regard to their medical information and data. However, HIPAA contains broad exceptions that provide medical administrators discretionary authority to disclose information to law-enforcement officers upon request. As a result, phenotypic testing of a bloodstain left at a crime scene that revealed that the suspect had a rare medical disorder could very well lead to a medical dragnet.
The European Union has issued a number of directives prohibiting the exchange of DNA information that can reveal physical traits. On June 25, 2001, a European Council resolution urged member states to exchange only the results of DNA analysis of noncoding chromosomal loci. Under the Prüm Convention of 2005, party states cannot make available DNA data drawn from coding segments of the DNA.56 Belgium and Germany explicitly prohibit deriving physical traits other than gender from DNA. In contrast, a 2003 amendment to the Dutch Code of Criminal Procedures permits investigators to obtain phenotypic information from DNA found at the scene of a crime when the suspect is not known.57 The Netherlands is the only country to explicitly approve the use of DNA forensic phenotyping.58
Increasingly, when law-enforcement investigators cannot get a match between the DNA found at a crime scene and the DNA profiles in their database, they use the services of specialized firms that construct phenotypic profiles of the suspected perpetrator from markers in the crime-scene DNA.59 This trend is likely to continue with the advent of gene chips, or DNA microarrays, such as those that have been developed by the company Affymetrix.60 These gene chips allow researchers to access information on thousands of genes simultaneously.
While at DNAPrint Genomics, Tony Frudakis predicted that in the future people’s DNA will be used to describe their physical appearance.61 Thus far there is little science to support that prophecy. Nonetheless, efforts to develop genotype-to-phenotype indicators will undoubtedly improve as financial investments in research, ancestry analysis, and forensic markets for crime-scene DNA profiles expand.
When a person’s DNA is taken for phenotyping, information about genetic predispositions can be revealed unexpectedly. For example, a suspect could learn the information for the first time in a courtroom or during sentencing. This scenario breaches a basic principle in medical law. People have the right to know—or not to know—about their genome. This right cannot be protected if forensic authorities gain access to sensitive genetic information and are authorized to use it as evidence against a suspect.
Some traits are less sensitive than others. If direct phenotyping of DNA could be limited to externally perceptible traits, such as hair color or stature, and nonsensitive internal or behavioral traits, such as voice type, left-handedness, or absolute pitch, then an appropriate balance between preserving genomic privacy and assisting law enforcement might be achieved. These traits raise no serious objections based on the right not to know, privacy, or the risk of stigmatization.62 Perhaps phenotyping the DNA from unknown subjects can fall under an ethical principle “the right not to know” when the proposed DNA test is for characteristics that are beyond the individual’s physical identifying features. Unfortunately, no such distinction has been made by crime investigators to date, and in the meanwhile, the temptation on the part of law enforcement to mine crime-scene DNA to make predictions about the physical, behavioral, or medical conditions of the alleged perpetrator only increases. Already claims have been made that genetic factors have been found that are associated with sexual orientation, intelligence, addictive behavior, and aggression. Even if they are unsound, law enforcement will be tempted to use them to generate profiles of suspects from the DNA, such as “Likely to be a tall, African American homosexual male, with high intelligence, a propensity for addiction, and recessive for sickle cell anemia.” Koops and Schellekens argue that “once phenotyping for criminal investigation is an accepted practice, phenotypical information will be used for other purposes as well, such as eugenics to redress bad genotypes related to aggression and pedophilia.”63