Summary
holding that "[c]linical errors [committed during DNA profiling] are far more likely to cause an inconclusive or no match result than a false positive."
Summary of this case from Taylor v. StateOpinion
January 9, 1992
Jeanne Mettler for defendant.
Carl A. Vergari, District Attorney (Barbara Egenhauser of counsel), for plaintiff.
The defendant, Dr. Morteza Mohit, an Iranian-born physician, was indicted for the rape and sexual abuse of a patient during the course of an office examination. Within a few hours of the incident samples for a sexual assault kit were collected, including a vaginal swab containing semen. The swab, together with blood samples from the victim and the defendant, were sent to the FBI Laboratory in Quantico, Virginia. DNA (deoxyribonucleic acid) was extracted from the semen and the blood samples and comparisons were made to determine whether the defendant's DNA "matched" the DNA extracted from the semen. It was determined that there was a "match." It was also determined that the probability of such a match occurring in the United States was 1 in 67,000,000 for Caucasians, 1 in 79,000,000 for Blacks, and 1 in 14,000,000 for Hispanics.
The defendant moved to exclude from trial the results of the DNA analysis, and a hearing was thereafter conducted to determine its admissibility.
As the use of DNA analysis in criminal cases is relatively new, there are no appellate decisions in the State of New York to provide guidance. In fact there are relatively few cases reported throughout the United States and they reflect a significant variety of opinion.
Standard for Admissibility
New York follows the legal standard of admissibility of scientific evidence originally set forth in Frye v United States ( 293 F 1013 [DC Cir 1923]). In this seminal decision, the court held: "Just when a scientific principle or discovery crosses the line between the experimental and demonstrable stages is difficult to define. Somewhere in this twilight zone the evidential force of the principle must be recognized, and while courts will go a long way in admitting expert testimony deduced from a well-recognized scientific principle or discovery, the thing from which the deduction is made must be sufficiently established to have gained general acceptance in the particular field in which it belongs." (Supra, at 1014; see, People v Hughes, 59 N.Y.2d 523, 537; People v Leone, 25 N.Y.2d 511; People v Allweiss, 48 N.Y.2d 40, 50.)
It was held in People v Middleton ( 54 N.Y.2d 42, 49) that "the test is not whether a particular procedure is unanimously indorsed by the scientific community, but whether it is generally acceptable as reliable." (See also, People v Smith, 63 N.Y.2d 41, 63.) There, in determining general acceptability of bite mark evidence, the court found that the techniques employed had been accepted and approved by a majority of experts in the field. Seizing upon this language, the Second Department in People v Bethune ( 105 A.D.2d 262, 267) cited a "majority" as establishing general acceptance of reliability. However, "counting heads" to determine a majority view is rarely feasible and can be of dubious value. (See, Andrews v State, 533 So.2d 841 [Fla App, 5th Dist 1988]; Caldwell v State, 260 Ga. 278, 393 S.E.2d 436.)
It seems unlikely that reference to a "majority" in Middleton (supra) was meant to establish a "majority" standard or rule. The scientific principles involved in this case were well settled and the decision simply rejected the notion of needing unanimous agreement within the scientific community. Most courts have not addressed what "generally accepted as reliable in the scientific community" actually means. While some degree of disagreement is inevitable, the amount of disagreement which will be tolerated has never been established. (See, Gianelli, The Admissibility of Novel Scientific Evidence; Frye v. United States, a Half-Century Later, 80 Colum L Rev 1197, 1201 [1980].)
It seems clear though, in principle, that if a well-respected minority within a given scientific community rejects as unreliable a particular procedure, technique, or theory, the court possesses the authority to agree with that minority view and exclude the evidence offered.
One of the few DNA cases reported in New York, People v Castro ( 144 Misc.2d 956, 959), sets forth a three-prong analysis to assist in the determination of the admissibility of DNA evidence:
Other New York cases include People v Wesley ( 140 Misc.2d 306 [1988]) and People v Shi Fu Huang ( 145 Misc.2d 513 [1989]), both dealing with the LifeCodes Laboratory.
I. Is there a theory, which is generally accepted in the scientific community, which supports the conclusion that DNA forensic testing can produce reliable results?
II. Are there techniques or experiments that currently exist that are capable of producing reliable results in DNA identification and which are generally accepted in the scientific community?
III. Did the testing laboratory perform the accepted scientific techniques in analyzing the forensic samples in this particular case? ( 144 Misc.2d, at 959.)
This analysis, however, says both too much and too little. If, for example, one looks at the second prong and ignores the first, nothing is lost (see, n 2, infra). In fact, at this point in time there is no serious dispute in the legal or scientific communities concerning the first prong. In Castro (supra) an error in the laboratory work left the issue of probability estimates moot, and this issue is the most controversial of all.
The third prong in Castro (supra) should go to the weight and not the admissibility of the evidence. If a laboratory uses a protocol which is generally accepted as reliable in the scientific community, errors in following the protocol or interpreting test results can be properly litigated in front of a jury. Once general acceptance in the scientific community is established, there is no benefit in having a Frye hearing on every case which follows.
Several factors should be understood when considering the admissibility of DNA evidence. First, there are a number of laboratories, public and private, which conduct DNA analysis, and their procedures differ. Second, the laboratory work will indicate whether or not there is a "match," but separate and apart from that is the question of calculating the probability of such a match. The laboratory work relies for the most part on principles of molecular biology, and the mathematical probability of a match relies on principles of human genetics and population genetics.
Within a given laboratory, for a number of reasons, procedures or protocols may change over a period of time. Within a given laboratory the sample populations, critical in determining probabilities, may also change over a period of time.
So a problem which remains is that even if we had decisions from the highest appellate courts, while helpful, they would only reflect on a particular laboratory, at a particular moment in time. Any subsequent changes in procedure or in the character of the sample population would invite new challenges.
It is well settled, however, that the court may find a scientific test reliable based on general acceptance as shown through legal writings and judicial opinions (see, Matter of Lahey v Kelly, 71 N.Y.2d 135, 144 [1987]). On this basis it may certainly be argued that the first prong under People v Castro ( 144 Misc.2d 956) is well established and need not be proven in future cases.
The hearings held in this case were extraordinary in that highly qualified experts for both sides were brought in from all parts of the country and they produced about 2,000 pages of testimony. When one considers the costs, in terms of time and money, for both the parties and the court, it is hard to imagine the same kind of process repeated in every DNA case which may arise. Unless there are uniform laboratory standards and procedures established which are deemed acceptable by statute, we will, in each case, in effect, be reinventing the wheel.
Be that as it may, the experience of this case allows this court to suggest the following analysis:
1. does the laboratory in question utilize procedures or protocols which are generally accepted as reliable within the scientific community (here the relevant scientific community being molecular biologists);
2. are the principles utilized by the laboratory in calculating the probability of a match generally accepted as reliable within the scientific community (here the relevant scientific community being population geneticists and human geneticists);
3. if the laboratory procedures are acceptable, but the probability estimates are not, is there a means of quantifying the probability of a match in a manner which would be generally accepted within the relevant scientific community, even if most would consider the estimate to be too high.
The purpose of this last analysis is to attempt to avoid prejudice to either side. If, as will be found in this case, a reliable match is made, but the probabilities attached are not reliable, should the proponent of the evidence be denied its admissibility altogether? Shouldn't the jury know that there was a match and that the possibility of the perpetrator being someone other than the defendant is remote, even if it is difficult to say precisely how remote? If, for example, many in the scientific community would agree that a probability is 1 in 1,000,000, but others reasonably doubting the accuracy of that number, can only agree to 1 in 100,000, shouldn't a jury at least know the more conservative number? The defendant could not reasonably claim prejudice, and the prosecution could still bring important and reliable evidence to a jury's attention. (For other cases on this issue see, Caldwell v State, supra [where the court reduced a LifeCodes estimate of 1 in 24,000,000 to 1 in 250,000]; Commonwealth v Curnin, 409 Mass. 218, 565 N.E.2d 440.) In Illinois v Fleming (Nos. 90-CR-2176 and 5546 [Cook County Cir Ct, Mar. 12, 1991]), the court rejected the proffered estimates and, refusing to set any other estimates, excluded the DNA evidence since it would have no evidentiary significance.
As will be discussed later in this decision, issues concerning population genetics are difficult to resolve, and particularly so in this case where the defendant comes from a distinct sub-group, not common in the United States.
In United States v Jakobetz ( 747 F. Supp. 250 [D Vt 1990]) the court left the issue of attacking a proper probability estimate to the jury. This approach appears to be patently improper. It is clear from Frye (supra) and its progeny that it is the court's role in the first instance to determine general acceptability in the scientific community. If the court cannot make that determination, the issue should not be presented to the jury.
* * *
I. LABORATORY PROCEDURES AND PROTOCOL A. DNA Identification and the FBI Protocol
Deoxyribonucleic acid (DNA) is a molecule containing genetic information which determines the physical characteristics of living organisms. DNA is contained in most cells of the body including blood, semen, saliva, and hair roots. Each of these cells have 23 pairs of chromosomes, each pair having one chromosome inherited from each of the parents. These chromosomes consist of the DNA which is composed of complementary strands attached in the form of a double helix, somewhat like a spiral staircase. The strands are attached at the rungs of the ladder by two of the four organic base pairs denoted Adenine (A), Thymine (T), Cytosine (C), and Guanine (G). A bonds only with T and C bonds only with G. The sequence of these bases enables DNA identification. Segments of the DNA strands, called VNTR's (Variable Number of Tandem Repeats), which are sequentially alike in all individuals, but vary in length, are analyzed in a typing procedure. These VNTR's are highly polymorphic and have unknown biological functions.
The FBI uses the process of Restriction Fragment Length Polymorphism (RFLP) analysis in measuring and comparing VNTR lengths in victim, suspect and evidence samples.
* * *
The results of the process are four autorads containing probed VNTR bands from specimens originating in the victim, suspect, and criminal evidence.
In comparing the size of the bands to each other, a match, no match, or inconclusive result is found. There are technical limitations to this science resulting in measurement imprecisions. Therefore, the FBI uses a 5% matching window. This matching window allows for the two bands being analyzed to be separated by plus or minus 2.5% to qualify under FBI criteria to be called a match. Since the possibility exists of others within the population having the same DNA profile, in order to give meaning to a match, probability estimates accounting for random matches must be made which take into consideration measurement imprecisions. A sizing ladder consisting of known fragment lengths is run along the autorad in every casework. These ladders serve as markers which form the structure of the binning procedure used for the FBI's population statistics.
B. Issues Relating to Reliability and General Acceptance in the Scientific Community
The defense contends that the laboratory results should not be admissible for the following reasons:
1. The use of Ethidium Bromide (EtBr) during gel electrophoresis causes band shifting and therefore does not produce reliable results.
2. The autorads in this case do not provide reliable results because:
(a) Two of the eight band comparisons had a differential of over 3% which occurs independently of each other only 3% of the time in the FBI database. For this to occur twice would produce a .09% probability, thereby justifying inconclusive or exclusive results.
(b) "Fat bands," occurring in this case, impedes the correct placement of the band, making the match a more subjective call.
(c) The sizing ladders are skewed by warping in the gel, further making a match call even more subjective.
(d) Three of eight of Dr. Mohit's blood sample bands match the corresponding three of the victim's blood sample bands. This so-called "coincidence" demonstrates the unreliability of DNA testing.
3. Clinical errors which can cause false positive results may occur as often as 1 or 2% of the time.
C. Protocol Arguments
* * *
D. Conclusions on FBI Protocol
1. Ethidium Bromide. There is no real evidence that the use of EtBr during electrophoresis causes unreliable results. Its use is generally accepted within the scientific community in both theory and practice. To the extent it may cause band shifting, the likelihood of causing a false positive over four probes is extremely unlikely. The possibility of distortions caused by EtBr may properly be argued before a jury, but the possibility of distortion does not affect the admissibility of the laboratory findings.
2. The issues of technical limitations of DNA typing leading to the problems of evaluating autorads may also be argued to a jury, but the reliability of the process is generally accepted within the scientific community.
3. Clinical errors are far more likely to cause an inconclusive or no match result than a false positive. Whether a clinical error was made in this case may be properly argued to a jury, but does not affect admissibility.
In conclusion, the FBI protocol for declaring matches employs techniques and procedures widely accepted and used in laboratories throughout the country, and is generally accepted as reliable in the relevant scientific community.
II. CALCULATING THE PROBABILITY OF A MATCH FBI Methods and Issues Raised
In order to provide meaning to a match, probability estimates must be provided to show how often the particular DNA profile occurs in a population. The estimates are derived by the FBI in the following manner:
A table of allele frequencies is made for the Caucasian, Black and Hispanic populations within the United States. This is accomplished by analyzing the DNA profiles of a sample database for each population. The resulting bands are grouped into bins. A bin represents a certain range of basepair lengths. Each bin is assigned a frequency dependent on what proportion of alleles from the database falls into each particular bin. The binning procedure provides conservative results according to the FBI. Each bin is larger than the 5% matching window so as to account for measurement imprecision. If a band lies on the border of two bins, it is assigned to the higher frequency bin. If a bin from the database contains less than five bands, it is merged together with the next bin.
Where a match is declared for a given case, multiplication rules of statistics are applied to arrive at an estimate for how often this DNA profile is expected to occur in the population. The first multiplication is based on what is called the 2PQ formula. For a given probe, two alleles will appear for a heterozygote on the autorad, one from each of the mother and father's chromosomes. A homozygote occurs when the same allele is inherited from the mother and father, resulting in only one band on the autorad. For a heterozygote, the allele frequency (represented as P and Q) of each of the two bins the alleles fall into are multiplied together. This number is multiplied by 2 because each allele has the possibility of occurring twice as there are two parents which could carry the allele. This process is performed on each of four probes. The four resulting genotype frequencies are multiplied together to arrive at the probability of a random match over four probes. This procedure was used for the Caucasian, Black, and Hispanic probability estimates claimed by the FBI in this case.
To justify the use of this method certain theoretical assumptions are relied upon. The defense in this case has attacked the validity of these assumptions and has demonstrated that within a certain segment of the scientific community these assumptions are not considered valid or reliable. The following issues are raised:
(a) The FBI improperly assumes Hardy-Weinberg and Linkage Equilibrium in computing the probability estimates.
(b) The FBI improperly assumes allele locus and loci frequencies occur independently of one another allowing multiplication rules of Hardy-Weinberg and Linkage Equilibrium to be utilized.
(c) The FBI's reference population does not account for varying genetic substructure in the various ethnic subgroups which could cause great deviations from Hardy-Weinberg Equilibrium.
(d) Inbreeding is a concern in this particular case which further emphasizes the problems of substructure and independence in the population.
(e) The reference sample used to determine allele frequencies is an inadequate representation of the reference population, as well as of the defendant's ethnicity. Further, it is insufficient in size to determine substructure in the population and to assess the probability of a rare event.
A. Hardy-Weinberg and Linkage Equilibrium
The first assumption is that the reference population is in what is called Hardy-Weinberg Equilibrium. This condition allows the use of the 2PQ equation and the four probe multiplication previously described. For Hardy-Weinberg to hold, certain further assumptions must be true. The population must be infinite in size, there must be random mating, and there can be no forces of evolution such as natural selection. All experts agree that no realistic population is ever in true Hardy-Weinberg Equilibrium. However, the FBI claims that adequate data on sample populations and use of the conservative binning method justify the assumption of random association of alleles, which in turn justifies the use of the multiplication rules without proving the condition of Hardy-Weinberg Equilibrium. The conventional test to show Hardy-Weinberg is to observe the expected amount of heterozygotes and homozygotes in the population which should exist if the population is truly in equilibrium. In their database studies reported in Fixed Bin Analysis for Statistical Evaluation of Continuous Distributions of Allelic Data from VNTR Loci, for Use in Forensic Comparisons (the Fixed Bin Paper), the FBI found an excess of homozygotes of the expected amount under Hardy-Weinberg conditions. However, the FBI claims this is due to technical problems which sometimes show only one band exhibited from a heterozygote. This can happen when a band is so small it runs right off the gel, or two bands occur so close to one another so as to appear as a single band. In casework, should a homozygote occur, a conservative method of multiplication is used. The point is moot here since all four probes show heterozygosity.
The FBI thus claims expected homozygosity and heterozygosity is not a sufficient test as to whether Hardy-Weinberg assumptions hold. The FBI claims in the Fixed Bin Paper that they have studied sample populations from North American Caucasians, Blacks, Hispanics, American Indians, Europeans, North Africans, Israelis, and Orientals, and there is significant randomness of alleles and random mating so as to justify probability estimates on the assumption of Hardy-Weinberg Equilibrium. The defense claims there is insufficient subpopulation data, and the data which does exist leads to contrary conclusions.
The FBI, in supporting its claim of Hardy-Weinberg, argues that mating is random since individuals are not aware of their partner's VNTR patterns. This argument is countered by population geneticists who claim that individuals mate according to race, ethnicity, physical appearance, as well as other attributes which could very well have a correlation with VNTR patterns.
The second assumption the FBI makes is Linkage Equilibrium. This allows the multiplication of the four genotype frequencies resulting from the Hardy-Weinberg multiplications. The final product of this multiplication is the statistical probability of a random match over all four probes. A requisite assumption underlying the multiplication is the Mendelian Law of Inheritance which says loci on separate chromosomes are inherited independently. If one assumes from this that there is no correlation between the genetic traits studied on each of the four chromosomes, multiplication is justified.
The primary issue is whether Hardy-Weinberg and Linkage Equilibrium hold and therefore justify the FBI's calculations. Sections B-E will focus on the major areas of controversy concerning the assumption of these conditions.
B. Independence
The issue of independence represents a major area of concern as it applies to the multiplication principles stemming from Hardy-Weinberg Equilibrium and Linkage Equilibrium. Independence assumed through Hardy-Weinberg Equilibrium allows the multiplication of the two allele frequencies at the specific loci of the genotype.
* * *
C. Reference Population
The FBI has chosen the three reference populations of Caucasians, Blacks and Hispanics on which they base their probability estimates. An assumption in using these reference populations is that there is no subheterogeneity resulting from the subpopulations in each group. That is to say, the alleles analyzed are sufficiently polymorphic among the various ethnic, racial and geographic populations to apply the over-all population frequencies for all individuals. Dr. Conneally explained that in order for substructure to occur in a population, two things must be true. The allele frequencies between subgroups must vary significantly, and there must be a high degree of nonrandom mating. The major argument against the assumption of homogeneity is insufficient evidence, and if this assumption proves false, great deviations from Hardy-Weinberg will occur.
* * *
Dr. Conneally worked out some numbers to show how even if substructure were apparent, the probabilities are still remote. He first pointed out that with substructure there should be a deficiency of heterozygotes, and therefore theoretically, the database would have rarer frequencies. Thus the probability estimates should be lower with substructure. He then explained that with substructure, we are essentially talking about error in allele frequencies. The chances are remote that all eight band frequencies would be wrong, and wrong by understating, rather than overstating the frequency. Chances are some will be higher than reality and some will be lower, and thus will mostly balance out. But for argument sake, he was asked to assume that all eight frequencies in this case were actually double the number reported. He then calculated the numbers and found a new estimate of 1 in 8 million (from 1 in 67 million). He then tripled the frequencies and came up with 1 in 2 1/2 million. These numbers show that even taking into account unrealistically large errors in allele frequencies, the probability estimates still reflect an extremely rare event.
D. Inbreeding
The issue of inbreeding is of particular importance in this case. The defendant, Dr. Mohit, was born in the Iranian Town of Shushtar. His ancestors over at least the past five generations were of Persian descent, and all from the same town or a town close by. They are all of the Shiite Muslim religion. Dr. Mohit claimed that for religious reasons, and as a matter of tradition, inbreeding was very common in his family. He indicated his maternal grandmother was the daughter of his father's great grandparents. Marriage among first cousins was common in his town.
In an article by Bittles, et al., of King's College, London, Reproductive Behavior and Health in Consanguineous Marriages, it was found that in the mainly Muslim countries, including those of Western Asia, "marriages contracted between persons who are related as 2nd cousins or closer account for between 20-55% of the total" (at 789). One must consider then the following from Mr. Waye's article, Forensic Analysis of Restriction Fragment Length Polymorphism: Theoretical and Practical Consideration for Design and Implementation, relating to his statement of binning as compensable for subpopulations, "The possible exception is inbred populations for which there may be an overall reduction in the number of alleles and a concomitant increase in the frequencies of individual alleles" (at 132). However, as just demonstrated, doubling and tripling frequencies while reducing probabilities still leaves extremely high numbers.
A study by Richard A. Nichols and David J. Balding of the University of London, Effects of Population Structure on DNA Fingerprint Analysis in Forensic Science, found that in the most extreme cases of inbreeding with massive uncle-niece mating, there was a correlation of 5%. Dr. Conneally explained this to mean a 5% deviation in the expected homozygosity under Hardy-Weinberg Equilibrium. They did find the highest correlation was tenfold less in European countries and more up-to-date data show considerable reductions due to modern transport. Dr. Budowle contended that although assuming the most extreme case would give more conservative numbers, they are not applicable as the defendant does not represent the extreme case. While it does appear the extreme case would apply more readily to Dr. Kidd's study of isolated South and Central American Indians than to individuals from the Middle East, it can be said that the defendant is not typical of the general United States population. He lies somewhere in the middleground. So the issue of independence of loci on chromosomes due to inbreeding also comes into play. Dr. Conneally worked out numbers for the court accounting for independence assuming an extreme case scenario of inbreeding. He claimed the highest degree of dependence between genotypes on separate chromosomes could not possibly exceed 10%. Factoring the 10% correlation into the multiplication of the four genotype frequencies, Conneally came up with a threefold more conservative estimate which factors out in this case to be 1 in 22 million. E. Reference Sample
The reference sample tested to determine allele frequencies of the larger reference population is of concern generally, and particularly in this case. There is significant controversy as to whether the FBI database sufficiently represents the population as a whole, and whether it would fairly represent someone of the defendant's background.
The need for the sample to be truly random, and thus, as close to a true representation of the reference population as possible, is essential for Hardy-Weinberg assumptions to be valid. Dr. Budowle has indicated that the FBI's procedure was to test their agents and accept their representation as to being Caucasian, Black or Hispanic. Dr. Mueller argues that the simple fact that the agents share the common thread of working for the FBI could mean the database is atypical of the general population. This refers to the possibility that certain segments of the population are prone to work for the FBI while others are not. Dr. Mueller further argues that although the FBI does not reveal this type of information, it is doubtful there are Shiite Muslims represented in the sample. By not inquiring further into the sampled agent's ethnic origins, it seems the attempt to use as random a sample as possible is not achieved.
Dr. Joel Cohen, of Rockefeller University, in his article DNA Fingerprinting for Forensic Identification: Potential Effects on Data Interpretation of Subpopulation Heterogeneity and Band Number Variability, appears to support the importance attached to this issue by Dr. Mueller.
Dr. Lander has contended that while it is difficult to find a truly random sample, there should be extensive questionnaires to achieve as diverse a sample as possible. Simply having individuals show up and claim to be either Caucasian, Black or Hispanic is insufficient.
* * *
F. Conclusions on FBI Probability Estimates
The evidence shows that there is sharp disagreement within the scientific community on the manner in which probability estimates are derived. It would appear that while human geneticists, on the whole, would find the FBI estimates acceptable, a significant number of respected population geneticists would not. The impression this court is left with, based on the record before it, is that human geneticists, more involved in the practical applications of genetics in dealing with disease, are not as concerned as the population geneticists in being more precise in citing probability estimates. More than one prosecution witness, for example, saw little relevance in being off by a power of 10. If the number is still very high, say 1 million instead of 100 million, what difference does it make? To the population geneticists, the difference is theoretically important.
It was quite revealing to hear Dr. Conneally say, while giving an example, "I want to make it clear to you, your Honor, that one in sixty million is not a precise number. I'm saying it's an estimate. It might be one in six million. It might be one in six hundred thousand." When pressed by the court to provide the most conservative possible estimate conceivable for eight bands over four probes, Dr. Conneally replied "one in one hundred thousand." He was not offended by the use of that number, it made no difference as far as he was concerned.
Does it matter in a criminal case if a jury is told 1 in 67,000,000 or 1 in 100,000? In most cases, probably not. But in a case where there is no reliable evidence other than the DNA evidence, it might mean a great deal. The difference in numbers might suggest that in the metropolitan New York area there could be 50 or more people who have a matching DNA profile, instead of, in theory, only 1 in the entire country.
The bottom line is that when speaking of probabilities in this context we are speaking of theories, not facts, in an area which is relatively new. There is still a great deal to be learned. As the size of databases grows over the years there is no question but that there will be significant changes in allele frequencies used to make computations. What the FBI reports as a 1 in 67,000,000 today, in a few years may be 1 in 670,000,000 or 1 in 6,700,000. It's hard to say. Further study on subgroups may reveal no significant differences or just the opposite. It may in time be generally accepted that no two people on earth will have the same DNA profile across four probes. However, the fact that it is difficult, given the present state of knowledge, to be precise, does not mean that conservative methods cannot be used.
If, in the interest of obtaining the most conservative of probability estimates, we accept the theory that there are significant differences between subgroups, and that the defendant in this case belongs to a subgroup representing a most extreme example of inbreeding, then factoring a correlation of 10% more than adequately compensates for any conceivable lack of independence, and increasing the frequencies reflected in the database can compensate for any conceivable deficiency in the database itself.
Based on the testimony of Dr. Conneally, as well as all the other evidence in this case, the court feels confident that no credible segment of the scientific community would claim that the probability estimates for Caucasians in this case or any other could be higher than 1 in 100,000. The vast majority would estimate a significantly lower probability. This court is of the opinion that by using the most conservative of all possible estimates, the prosecution may adequately demonstrate how unusual it would be for someone else to have the same DNA profile, while at the same time avoiding any possible prejudice to the defendant.
By reducing in the same proportion, the estimates for Blacks would be 1 in 118,000 and for Hispanics, 1 in 21,000.
CONCLUSION
For the reasons discussed herein the defendant's motion to suppress DNA evidence is denied; however, the FBI probability estimates shall be limited in accordance with this court's decision.
[Portions of opinion omitted for purposes of publication.]