Opinion
Case No.: 12-CR-0792 YGR
05-02-2016
UNITED STATES OF AMERICA Plaintiff, v. HENRY CERVANTES, et al., Defendants.
TRIAL ORDER NO. 2 REGARDING LOW COPY NUMBER DNA TESTING AND STATISTICAL ANALYSES
RE: DKT. NO. 895
Pending before the Court is Jaime Cervantes's ("J. Cervantes") motion in limine to exclude low copy number DNA testing conclusions and weak statistics, and request for a Daubert hearing (Docket No. 895). At issue in this motion are five conclusions regarding DNA mixtures found on: (1) a Nike shoe; (2) black zip ties; (3) a pry bar; (4) other zip ties; and (5) a passenger side car door panel. (See id. at 2.) J. Cervantes withdrew an initial request for a Daubert hearing, but confirmed at the March 18, 2016 hearing that he still moves to exclude the five conclusions as unreliable, without probative value, and unduly prejudicial. (See Docket No. 917.) As discussed below, the Court declines to exclude the evidence on the current record, but requires additional information to assess reliability in terms of the methodology used in relation to what J. Cervantes identifies as tests of DNA mixtures and as having low input amounts. Separately, under Rule 403, the Court will exclude four of five analyses that resulted in statistical random match probabilities of "one in four," "one in two," and "two in three," (see Docket No. 895 at 5-7), without prejudice to a motion to admit if J. Cervantes suggests to the jury that DNA analysis should have been done on those items and was not.
I. Low Copy Number DNA Testing and Mixed Samples
As a preliminary matter, the government correctly notes that J. Cervantes filed his motion on January 20, 2016, after this Court's deadline for Daubert motions and an October 13, 2015 Daubert hearing on DNA and other expert-related issues. (See Docket No. 938 at 1-2; Docket No. 755.) At a March 18, 2016 hearing, J. Cervantes explained that an expert with whom he consulted did not finish an evaluation of the government's DNA testing until January. J. Cervantes does not otherwise explain the delay in bringing this motion. J. Cervantes's delay prevented the Court from addressing low copy number issues at a hearing in October when the government's DNA expert testified regarding statistical—but not low copy number—issues.
J. Cervantes stated at the hearing that his expert did not write a report, and is not offering the expert at trial.
Notwithstanding the significant delay, the Court must perform its gatekeeping function and determine whether the methodology at issue is reliable before the proffered testimony at issue could be presented to the jury. See Daubert v. Merrell Dow Pharms., Inc., 509 U.S. 579, 592-97 (1993); Estate of Barabin v. AstenJohnson, Inc., 740 F.3d 457, 463-64 (9th Cir. 2014) (en banc); Fed. R. Evid. 104. As this Court has explained, "[i]n federal courts, the admission of expert testimony is governed by Federal Rule of Evidence 702, as elucidated by the Supreme Court in Daubert," Barabin v. AstenJohnson, Inc., 700 F.3d 428, 432 (9th Cir. 2012). Federal Rule of Evidence 702 allows expert testimony only if the expert's "scientific, technical, or other specialized knowledge will help the trier of fact to understand the evidence or to determine a fact in issue." Fed. R. Evid. 702. Rule 702 permits experts to testify if their testimony is: (1) "based upon sufficient facts or data," (2) "the product of reliable principles and methods," and (3) the result of applying those principles and methods reliably to the facts of the case. Id. In determining whether an expert's testimony meets the standards of Rule 702, the court acts as a "gatekeep[er]" that "ensur[es] that [the] expert's testimony both rests on a reliable foundation and is relevant to the task at hand." Daubert, 509 U.S. at 597; see also Kumho Tire Co. v. Carmichael, 526 U.S. 137, 148-49 (1999). In addition, expert testimony that is "otherwise admissible may be excluded under Rule 403 if its probative value is substantially outweighed by the danger of unfair prejudice, confusion of the issues, or misleading the jury." U.S. v. Rincon, 28 F.3d 921, 923 (9th Cir. 1994) (citing Daubert, 509 U.S. at 595).
The Court needs additional information, particularly in three areas set out below, to determine whether the methodology used is reliable. See Estate of Barabin, 740 F.3d at 463-64. Cf. United States v. McCluskey, 954 F. Supp. 2d 1224, 1276-77 (D.N.M. 2013) (addressing reliability of low copy number DNA testing, as distinct from standard testing methodologies and statistical analyses). As a general matter, J. Cervantes bases his reliability concerns on the SERI laboratory and Virginia Sadl's analyses of mixed samples (samples containing DNA from multiple sources) with small amounts of DNA (measured in picograms) and the documented appearance of stochastic effects. (See Docket No. 895 at 3-7.) The government asserts that J. Cervantes's arguments do not apply because, here, Sadl used the "Standard Method," not a low copy number method. (See Docket No. 938 at 3, 5-6.) J. Cervantes explains, however, that the government misses his threshold point: "In its simplest usage, [low copy number] refers to the amplification of a sample in which there is insufficient DNA to ensure reliable testing that is free of stochastic effects." (Docket No. 966 at 1-2; see also United States v. Williams, 2009 WL 1704986, at *3 (C.D. Cal. June 17, 2009) ("The weight of authority supports [the] contention that [low copy number] testing can occur without an increase in the PCR cycle number.").)
The court in McCluskey considered the Daubert factors in concluding that (1) a lab testing DNA at input amounts below its own input quantity threshold for reliability failed to demonstrate that the lab "[wa]s able to obtain reliable DNA profiles from samples below [that amount,] 250 [picograms], or to reliably interpret such profiles[,]" id. at 1288; (2) "peer review and publications have raised serious questions about the reliability of testing low amounts of DNA and accounting for stochastic effects[,]" id.; (3) "the recognition of stochastic effects constitutes acknowledgment of a significant risk of error[,]" id.; (4) "there are standards controlling [low copy number] testing, to the extent that each laboratory is required to empirically establish its own stochastic threshold[,]" id.; and (5) "the reliability of [low copy number] testing is not 'generally accepted in the relevant scientific community[,]'" id. (quoting Daubert, 509 U.S. at 593-94). Then, the court considered the "foundational matter" of showing that a DNA sample from an individual (one source) exceeds the minimum picogram amount, rather than allowing a sample with multiple sources to avoid being categorized as low copy number by aggregating the amounts from all sources to determine a picogram amount ("mixed samples"). See id. at 1288-90.
First, in particular, the Court needs more information regarding Sadl's application of standard techniques to low quantities of DNA and mixtures of DNA, and whether or how Sadl accounts for the appearance of stochastic effects in these circumstances. As noted in cases both parties cite, there is a meaningful distinction between laboratories that incorporate previously validated specialized techniques for low quantity samples, see United States v. Morgan, 53 F. Supp. 3d 732, 735-39, 741-42 (S.D.N.Y. 2014), and laboratories that use "no special procedures or methods of interpretation for [low copy number] testing," McCluskey, 954 F. Supp. 2d at 1277. "Other laboratories testing [low copy number] samples modify their protocol, apply different interpretive principles to account for stochastic results, or perform replicate testing—but the [lab in McCluskey] applie[d] the same protocol for non-[low copy number] testing to [low copy number] samples, despite recognizing that there are different problems in [low copy number] testing and different challenges in formulating reliable interpretations." Id. at 1286 (emphasis added). Sadl suggests that SERI fits into the McCluskey category, not the Morgan category: laboratories that apply standard techniques to low quantities of DNA. (See Docket No. 938-1 ¶ 6.) The Court lacks a basis to find that doing so was reliable for the DNA quantities at issue here. Cf. McCluskey, 954 F. Supp. 2d at 1277-81 (noting a meaningful distinction in case law and scientific literature between reliably employing standard techniques for recommended quantities of DNA and employing standard techniques for low quantities of DNA).
See also Morgan, 53 F. Supp. 3d at 736 n.2 (explaining that McCluskey "contrasted [the lab in McCluskey] with [the lab in Morgan], noting that [the latter] 'has done extensive internal validation of its [low copy number] testing and has received certification and approval for it . . . .'" (citation omitted)).
Second, the Court needs more information about the quantity of DNA from each source tested in each sample. As the court in McCluskey observed about the scientific literature:
Different definitions [of low copy number testing] focus on: (1) the amount of DNA tested (often setting a limit at less than 200 pg or less than 100 pg); (2) modifications to methodology to increase sensitivity (often increasing the number of PCR cycles from 28 to 31 or 34); or (3) DNA profile appearance exhibiting increased imbalance of observed alleles. "In all these definitional approaches, it is recognized that data reliability is inferior when lower amounts of DNA are tested and thus additional measures must be taken to improve the chance of obtaining results that accurately reflect the sample being examined." Another recognized authority states that "LCN typing is better defined as the analysis of any results below the stochastic threshold
for normal interpretation." "The success rate is low; often the results cannot be interpreted or are meaningless for the case."McCluskey, 954 F. Supp. 2d at 1277-78 (citations omitted); see also id. at 1279 (explaining that, independent of whether the polymerase chain reaction/short tandem repeat methodology could be reliable, "there are few reported cases on admissibility of [low copy number] testing under Daubert and Rule 702, and the scientific literature is unclear and often addresses different procedures").
Here, Sadl has not indicated whether or how she accounted for the issues J. Cervantes raises (i) that DNA for some items constituted mixtures from multiple potential contributors and (ii) that the DNA amounts fell outside a reliable per-individual picogram range. Cf. McCluskey, 954 F. Supp. at 1290 (explaining the need to determine not the total quantity of DNA in a sample, but the quantity contributed by an individual). For instance, J. Cervantes notes that fewer than 100 picograms were amplified for the Nike shoe. (See Docket No. 895 at 5.) Sadl indicates that she "followed Standard Methods of testing, which include standard amplification cycles of 28 or 29 cycles as recommended by the manufacturer of the Identifiler Plus kit, approved by the FBI, and validated by SERI, and no post-amplification procedures." (Docket No. 938-1 ¶ 6.) Yet her declaration does not (i) indicate the amount of input DNA for each sample, (ii) indicate the amount of input DNA for each source within each sample, (iii) explain under what particular protocol she could test items with, for example, fewer than 100 picograms of input DNA, or (iii) provide pertinent sections of the "AmpFlSTR® Identifiler® PCR Amplification Kit User's Manual" for the kit used in this case to know how many amplification cycles should be used for reliable testing according to Applied Biosystems's validation studies, cf. Docket No. 691-2 (including a portion of the User's Manual regarding population data).
SERI protocol 6.6.1 ("Strong Samples") states, in part: "Caution must be taken with regard to the concentration of input DNA added to the PCR reaction utilizing the Minifiler or Identifiler Plus kits. The amount of input DNA should fall between a range of 0.25 and 0.5 ng [250 and 500 picograms]. (volume 10ul)." (Docket No. 1032-1 at 14 (emphasis added).) SERI protocol 6.6.2 ("Weak Samples") states:
When the total number of allele copies added to the PCR is extremely low, e.g., less than 35 pg, unbalanced amplification of the two alleles of a heterozygous individual may occur. This is due to stochastic fluctuation in the ratio of the two different alleles. The PCR cycle number and amplification conditions have been specified to produce peak heights of less than 150 RFU for a sample containing 35 pg human genomic DNA (corresponding to ten total allele copies).(See Docket No. 1032-1 at 14.)
If peaks are below 150 RFU, they will not be recognized if analyzed at 150 RFU. To obtain higher peaks, samples can be reinjected using increased PCR product and/or injection times. Alternatively, determine the height of the lowest obvious peak and reanalyze at a lower RFU but not below 50 RFU. (Peak heights <150 RFU should be interpreted with caution).
J. Cervantes cites scientific literature that the precise number of cycles may correspond to the appearance of stochastic effects. (See Docket No. 895-3, Ex. C ("When present in low copy number, a molecule that is amplified by chance during the early rounds of the PCR is likely to be preferentially amplified. . . .").)
Third, aside from DNA input amounts, J. Cervantes identifies stochastic effects in the analyses at issue in this motion, though the Court lacks information as to (i) whether or how Sadl addressed them and (ii) whether Sadl performed replicate or comparative analyses. Cf. McCluskey, 954 F. Supp. 2d at 1285 (citing evidence "that one of the criticisms of [low copy number] interpretation is that, once stochastic effects have occurred at some loci, there is no way to be certain that the stochastic effects were confined to those loci and did not occur at other loci"). Instead, Sadl states that "[e]very analysis exhibits some stochastic effects, regardless of DNA input amount," (Docket No. 938-1 ¶ 8), and that, in this case, "any potential stochastic effects were considered and accounted for in the interpretation of the mixtures in these samples," (id.; see also id. ("With a . . . low amount of DNA one observes slightly more stochastic effects. That is why laboratories establish a stochastic threshold and determine peak height ratios during validation to use as a tool to determine when drop out events have occurred.").). These conclusory opinions do not illumine precisely how specific "potential stochastic effects" were "considered and accounted for" in Sadl's interpretation. Rather, the acknowledgment of stochastic effects in testing a sample raises unanswered questions about the reliability of testing that sample as a whole. See McCluskey, 954 F. Supp. 2d at 1283-86; Docket No. 966 at 2; Docket No. 895 at 5-7; see also McCluskey, 954 F. Supp. 2d at 1284 (noting that "authoritative articles state that [replicate testing] is an important means of ensuring reliability for [low copy number] testing").
As a separate matter, in testing the Nike shoe, Sadl explains that some alleles were above her lab's stochastic threshold and others below it. (See Docket No. 938-1 ¶ 9.) Sadl quotes part of the SWGDAM Guidelines. SWGDAM Guideline 4.6.3.2 states in full: "A restricted CPE/CPI may be applied to multiple major contributors despite the presence of minor contributor(s) alleles below the stochastic threshold; a description of how to calculate can be found in Section 5.3.5." 2010 Scientific Working Group on DNA Analysis Methods ("SWGDAM") Interpretation Guidelines for Autosomal STR Typing, available at https://www.fbi.gov/about-us/lab/biometric-analysis/codis/swgdam-interpretation-guidelines. Next, Section 5.3.5.3 states: "Restricted CPI and CPE. Given (a) a mixture at a locus having alleles P, Q, R, and S, (b) alleles P, Q, and R significantly (as defined by the laboratory) above the stochastic threshold, and (c) allele S is below the stochastic threshold, the interpretation might be that the higher RFU alleles are a distinct group, separate from the contributor(s) of the low-RFU S allele. The lab might choose to calculate a restricted probability of inclusion utilizing just the P, Q, and R alleles, (p + q + r)2." Id. (emphasis added). Here, however, J. Cervantes argues that two peaks associated with him are "just barely"—not significantly—above the stochastic threshold (at 153 RFU), thus, potentially affecting any reliability determination. (See Docket No. 895 at 5.)
In sum, the Court lacks sufficient information to make a reliability determination regarding tests that J. Cervantes argues are low copy number and exhibited stochastic effects.
II. Rule 403 and Random Match Probabilities
Separately, J. Cervantes moves to exclude DNA testing on the black zip ties, pry bar, additional zip ties, and interior passenger side door panel under Rule 403. The probability that a randomly selected unrelated individual would be included as a contributor for each of these items ranges from "one in four" to "one in two" to "two in three." (See Docket No. 895 at 5-6.)
The Court will exclude the testing on these items under Rule 403, without prejudice to a motion to admit if J. Cervantes suggests to the jury that DNA analysis should have been done on these items and was not. The Court agrees with the government that there is no bright line for statistical significance. Nonetheless, here, exclusion of the testing resulting for these four random match probabilities is appropriate under a Rule 403 analysis. The government's argument that the evidence is probative of the fact that J. Cervantes "is a possible contributor," (see Docket No. 938 at 7-8), is minimal at best and risks undue prejudice and confusing the jury with statistical data by showing a possibility of inclusion, but at the same time, showing a high likelihood that an unrelated individual in the population selected at random also would be included. The balance may change if the defense opens the door.
For example, Sadl herself indicates that a "one in two statistic" means that "it is extremely common to be a possible contributor." (Docket No. 938-1 ¶ 11.) The same goes for the "one in four" and "two in three" statistics. And Sadl indicated that she must conduct statistical analyses in every case as a matter of protocol, not that doing so is particularly probative for any particular test. (See id.) --------
III. Conclusion
For the reasons above, the Court ORDERS that the government not present opinion testimony regarding the probability of J. Cervantes's DNA appearing on the Nike shoe without further hearing on the topic. The Court GRANTS WITHOUT PREJUDICE J. Cervantes's motion to exclude testimony regarding the other four items—black zip ties, pry bar, additional zip ties, and interior passenger side door panel—under Rule 403.
IT IS SO ORDERED. Date: May 2, 2016
/s/_________
YVONNE GONZALEZ ROGERS
UNITED STATES DISTRICT COURT JUDGE