Table 1.2 also shows the frequencies with which each specific short tandem DNA segment would be observed in 13 loci (26 alleles) for a hypothetical reference population. Some geneticists like L. A. Zhivotovsky argue that the reference population is critical in establishing the probability of a random match between two samples because the frequencies of matched STRs are likely to be higher and the random-match probability also higher for people in the same geographical region or of the same ethnic or racial ancestry.9 If the suspect is Asian, the chance of finding a random match in the European population is less than it would be for the Asian population. Because the suspect is presumed innocent, it is incumbent on our criminal justice system to consider seriously the conditions under which someone else in the population who has a forensic DNA profile identical to that of the suspect left his or her DNA at the crime scene. For locus A, the frequency of 10/6 STRs is .01, or 1 in 100. Similarly, for loci B, C, and D, the frequencies within the reference population are .06, .03, and .08, respectively.
TABLE 1.2 Calculating the Probability of a Match in Two DNA Samples for 13 Loci
The next step is to determine the rarity of the forensic DNA profile in the population. If the STRs at the loci are truly independent (as they are assumed to be), the chance that two people in a reference population will have the same number of repeats is obtained by taking the product of the frequencies for each locus. This is analogous to determining the chance of getting two heads when one simultaneously flips two different coins (which are independent events). The chance of getting a head on the first flip is .5, and similarly for the second. The chance of getting two heads on flipping both coins is .5 × .5 = .25. Thus, if one flips both coins enough times, then one-quarter of the simultaneous flips will yield two heads.
For the first four loci in our example, the frequency of the specific STRs in a reference population would be determined by (A = .01) × (B = .06) × (C = .03) × (D = .08) = 1.44 × 10−6, or about one chance in a million people. Thus in a population of 6 million people, approximately 6 people would be expected to have the same eight STRs on four loci.
If a crime were committed on an isolated island that had only 500 people, then a random-match probability of one person in nearly 1 million might be high enough to convince a jury that the suspect is the source of the crime-scene DNA, but few places are that isolated. If we completed the calculations for all 26 alleles, the probability of getting a random match in STRs for two people would be quite small—typically less than one in many billion.
Criminal justice agencies may differ in what they choose as the match probability that eliminates any doubt they have that a suspect was the source of the crime-scene DNA. Whatever number they choose as evidence of a “definitive match,” they should not neglect to consider that there is a much greater likelihood that a close relative would have a matching profile.
The Laboratory Process for DNA Analysis
When evidence is received by a facility, it is characterized and coded, and a determination is made whether it is appropriate for DNA testing (see figure 1.8). If so, it is sent to a section of the laboratory for DNA extraction. Whether a garment, blood, hair, or saliva is involved, the extraction process removes proteins and other non-DNA components, leaving purified DNA residue for analysis. The extraction process may also involve the separation of sperm-cell and non-sperm-cell mixtures, such as sperm cells from vaginal epithelial cells. Different methods and reagents are used for the extraction of DNA depending on the nature and homogeneity of the sample. Once the DNA is extracted and purified from the sample, it is sent for DNA quantification, a process of measuring and setting the proper amount of DNA so it falls within the optimum ranges for analysis.
As an example, in a rape case the sample might consist of a stain, which is expected to contain a mixture of the perpetrator’s sperm and the victim’s epithelial cells (cells on the outer layer of the skin or organs). A portion of the sample is removed and exposed to a group of chemical reagents that degrade the proteins and break open the epithelial cells to release the nuclear DNA. The sperm pellets remain intact, and the freed epithelial DNA in solution is removed and analyzed. The original sample has undergone “differential extraction” when sperm and nonsperm fractions of the sample have been separated. The sperm pellets can then be exposed to another set of reagents to isolate the pure DNA. The DNA analyzer is optimally designed to handle certain amounts of DNA. If the DNA quantities are too large, artifacts (any result which is caused by the DNA analyzer itself and not by the DNA entity being analyzed) and other problems will likely occur, such as split peaks (a peak is an indicator of an allele at a locus) or off-scale peaks. If there is too little DNA, then some peaks will not appear. This is called allele dropout or possibly locus dropout. The analysis of STRs works best for samples containing 0.25 to 2.0 ng of DNA. One nanogram of DNA is contained in about 160 to 168 cells.
When the profiling is complete, the DNA analyzer produces a graphic and numerical output (an electropherogram) to a computer. The output shows a continuous oscillating line with periodic peaks and numerical values next to the peaks. The heights of the peaks represent the amount of DNA that was amplified for each allele segment (see figure 1.9), which is related to the amount of evidentiary DNA used for the analysis.
FIGURE 1.8. Flow diagram of how DNA evidence is handled in a forensic laboratory. Source: U.S. Department of Justice, http://www.justice.gov/oig/special/0405/chapter2.htm (accessed May 23, 2010).
If the DNA comes from a single individual and is not degraded, two peaks should appear for each locus where the individual has inherited different alleles from each of his or her parents (heterozygous for that locus). In cases where an individual has inherited the same allele from each of his or her parents (homozygous for that locus) only one peak will appear (figure 1.9). If the DNA consists of a mixture of two individuals, there can be anywhere from one to four peaks at a particular locus, significantly complicating the picture. For example, if both individuals are heterozygous at a particular locus and have the same STR alleles at that locus, only two peaks will appear; if the two heterozygous individuals have no STR alleles in common, then four peaks will be observed; and if they share the same STRs for just one allele, then three peaks will be observed.
FIGURE 1.9. The output from a forensic DNA analyzer (called an electropherogram) showing the STR alleles for two contributors at three loci. The top number reports the number of STR repeats for the allele. The bottom number is a measure of the height of the peak (in RFUs), which represents the quantity of DNA (vertical axis). Source: Dan Krane.
The position of the peaks on the x-axis of the electropherogram indicates how long it takes the allele to pass through a capillary tube, which correlates with the length of the DNA segment, and therefore indicates the number of repeats. The height of the peaks in “relative fluorescence units” (RFUs) measures the amount of DNA present in the sample. Peaks from all the alleles in a profile coming from the same individual are expected to have about the same heights. The computer-generated numbers under each peak tell us how many repeats there are in the sample at that locus and the height of the peak relative to a baseline.
Quality-Control Procedures
The machines are calibrated and the reliability of the overall assay is monitored by analysis of samples of known DNA type as a control standard (like an official measuring rod). In all accredited forensic laboratories and others that follow good scientific practice, technicians are instructed to include a sample of known DNA sequence in every analysis sample they undertake. Any analysis where systematic errors have occurred (affected possibly by temperature, pressure, vibrations, or machine malfunctions) will be reflected in the control sample (see chapter 16). In addition, every step in the process, including technician interpretations, is required to be documented for purposes of troubleshooting and replication.
For example, the New York City forensic DNA lab has designed a number of control procedures that aim to minimize the possibil
ity of false positives (false matches) and false negatives (false exclusions). The facility has an isolated laboratory that tests all reagents for purity and for any traces of DNA. The reagents are put through a DNA analyzer to validate that they are DNA free. Every sample of DNA that is put through the DNA analyzer is required to be checked against a DNA sample that has a known STR profile. If the profile of the quality-control standard does not match its expected values, the run is invalidated. There are strict protocols for how evidence moves through the different stages in the testing regime: The people who initially prepare the evidence for testing are different from those who do the extractions or place the sample in a DNA analyzer. The stages in the DNA analysis are segmented by physical spaces and are also separated by personnel who conduct different tasks in the process. In addition, those who handle the evidence during extraction, the PCR, or DNA analysis are required to be suited up from head to toe in protective garments designed to prevent cross-contamination of DNA. No one whose DNA is not on file is allowed to enter the high-security laboratory. In case there is a breach in the protective outerwear, the contaminant DNA from the technician can be traced.
Notwithstanding the strict quality-control procedures adopted by some laboratories and the crisp-looking output of electropherograms, there is still room for judgment calls in reading the output of the analyzer. In the words of William Thompson and colleagues, “Although many technical artifacts are clearly identifiable, standards for determining whether a peak is a true peak or a technical artifact are often rather subjective, leaving room for disagreement among experts.”10
It is the policy of the FBI’s forensic DNA laboratory that it will not report out a DNA profile with fewer than 10 complete STR markers. The New York City lab’s policy is less stringent; it will accept 6 or more good STR markers. Also, the New York City lab’s policy is that it will not enter a single peak at a locus (representing one allele) if the second peak at the locus (representing a second allele) is known to have been badly degraded and not found in the output. The computer program has a peak (height) threshold for scoring alleles. If the peak height is below 75 RFU, it is not reported. This avoids giving the forensic DNA analyst some discretion in calling the peak real or spurious. By the time it reaches 75 RFU, the peak is considered unambiguous. Other laboratories may use a different threshold below which peaks are ignored. Erin Murphy observes: “The FBI Protocol allows only peaks over 200 relative fluorescence units (RFU) to be considered conclusive for match purposes, though it recommends interpretations of any peak over 50 RFU including for exculpatory purposes.”11 Thus the FBI has ostensibly adopted 50 RFU as the detection threshold.
DNA analysis for use in criminal investigation is a highly complex process that involves collecting biological samples from individuals and crime scenes, extracting DNA from those samples, using the PCR to amplify the amount of DNA from subanalytical levels to analytical levels along noncoding regions of the DNA, and then using some method to determine the genetic variation of the sample. The result of this process is a DNA profile that can be stored electronically and used to identify an individual or to link that individual to a crime or exclude him or her from it by comparing that profile with others.
A new generation of DNA analyzers, genetic typing techniques, and forensic laboratory protocols has dramatically increased the efficiency of DNA testing and has helped address the most obvious sources of error of the first generation of DNA testing systems. Separation of samples, isolation of functions, proficiency testing, and protocols for interpreting mixed samples are among some of the quality-control innovations practiced, in theory, by all accredited laboratories. The shift from the autoradiograph with the dark bands generated by RFLP and gel electrophoresis to the PCR and the electropherogram has also reduced the likelihood of reading errors.
Nonetheless, no technology is errorproof, and DNA analysis is no exception to this rule. Artifacts in the output can be produced by the machine because of imperfections or environmental effects. Human error can also occur. Sample contamination, misinterpretations or mischaracterizations of data, and the reporting of inaccurate or misleading statistics are all sources of human error that can and have occurred in DNA cases, sometimes resulting in the wrongful conviction of an innocent person. The potential for error and abuse was made painfully clear in the case of the infamous Houston Police Department Crime Laboratory, where a final report by an independent investigator revealed that, of a total of 135 DNA cases reviewed, major issues were identified in 43—or approximately 32 percent—of those cases, including four death penalty cases. The report found that:
The Crime Lab’s historical DNA casework reflects a wide range of serious problems ranging from poor documentation to serious analytical and interpretive errors that resulted in highly questionable results being reported by the Lab. The profound weaknesses and flawed practices that were prevalent in the Crime Lab’s DNA work include the absence of a quality assurance program, inadequately trained analysts, poor analytical technique, incorrect interpretations of data, the characterizing of results as “inconclusive” when that was not the case, and the lack of meaningful and competent technical reviews. Furthermore, the potential for the Crime Lab’s analysis of biological evidence to result in a miscarriage of justice was amplified exponentially by the Lab’s reported conclusions, frequently accompanied by inaccurate and misleading statistics that often suggested a strength of association between a suspect and the evidence that simply was not supported by the analyst’s actual DNA results.12
The fallibility of DNA testing will be discussed in more detail in chapter 16.
Fallible or not, DNA is here to stay. Few would argue with Simon Cole’s statement about the stature of forensic DNA in criminal justice: “New technology is changing the administration of criminal justice. Among the most prominent of such changes is the development of forensic DNA technology, which includes a forensic assay with potentially enormous discrimination and sensitivity and the development of large databases based on that assay.”13 However, as stated in the book Truth Machine, “This seemingly unassailable, transcendent form of criminal evidence remains bound up with stories that are infused with contingent judgments about the mundane meaning and significance of evidence.”14
Chapter 2
The Network of U.S. DNA Data Banks
The purpose of DNA databases is to solve crimes that would otherwise be unsolvable.
—John M. Butler1
The power to assemble a permanent national DNA database of all offenders who have committed any of the crimes listed [in 18 U.S.C.] has catastrophic potential. If placed in the hands of an administration that chooses to “exalt order at the cost of liberty” . . . the database could be used to repress dissent or, quite literally, to eliminate political opposition.
—Judge Stephen R. Reinhardt in
United States v. Kincade (2004)2
In their original conception, forensic DNA data banks were designed to hold DNA profiles of violent felons and recidivist sex offenders. Because of their initial successes, over the last 15 years there has been an inexorable drive among law-enforcement agencies to push their use to their limit, stopping just short, at least for the present, of including every person’s DNA in a national network.
By the new millennium every state law-enforcement agency and many local ones in the United States had become connected through a coordinated and federally managed forensic DNA data bank. At the heart of this system is a computer network overseen by the Federal Bureau of Investigation called the Combined DNA Index System, or CODIS. This network provides police with tools to solve “cold cases,” that is, cases that have been dormant because of lack of leads, to link a current suspect to a crime scene, and to track suspects who may seek to change their identity.
The Origins and Workings of the Combined DNA Index System (CODIS)
CODIS was initiated in 1990 as a pilot program serving 14 state and local laboratories. The main idea behind the program was to collec
t, analyze, and store the DNA of individuals convicted of felony sex offenses and other violent crimes on the theory that these offenders were likely to be recidivists and frequently leave biological evidence at the scene of a crime.3 By maintaining their DNA on file in a shared database system, police could develop investigative leads and solve crimes that might have been committed in the past or would be committed in the future by the same individual both within and across jurisdictions.
Genetic Justice Page 4