strmix vi.0.7.49 truealiele casework purpose...• strmix is a stand-alone program written in java...

STRmix V I.0.7.49 TrueAliele Casework

Purpose

Casework Interpretation of Complex DNA Mixtures PART I: Differential degradation studies

December 10,2014

To examine the sensitivity and reproducibility of STRmix and TrueAliele Casework when testing challenging samples. Part I of this study focuses on the following sample types: 2-person and 3-person mixtures created inhouse that have differential degradation as an element of comp lexity.

Methodology a nd Results

System versions and settings

STRmix (ESR) • STRmix is a stand-alone program written in Java (Oracle) for Windows. Prior to use, laboratory-specific

parameters were established for the Identifiler Plus/3130XL combination of multiplex amplification kit and detection platform. These parameters included a regression analysis of inter-allelic stutter levels at each locus, the detection system's signal saturation point, probability of drop-in, allele amplification variance for a given template quantity, and inter-locus amp lification variance. The first two parameters were assessed outside of the software with Excel. The latter two parameters were assessed through the STRmix Model Maker module and, with additional calculations in Excel for Model Maker version 2.0.

• Interpretations were performed using version 1.0.7.49. • Model Maker version 1.0.7.49 was used to establish the inter-locus amp variance. • Unless noted otherwise, studies were tested using two separate allele amplification variance settings. One

was from the Cal DOJ analysis using Model Maker version 1.0.7.49. The other was calculated for Cal DOJ by ESR using the Model Maker approach incorporated into the commercial version 2.0.

TrueAliele Casework (Cybergenetics) • TrueA liele Casework is a software and hardware system. The software employs the MATLAB

programming language. Hardware includes a Linux-based server with a Macintosh interface. Additional PC workstations have been added to the system. All settings were established by Cybergenetics with reference to a set of sample files supplied by Cal DOJ .

• Analyses for this study were performed using an HP ProLiant ML350 G6 with 8 parallel processors; server version 3.25.444. 1; VUler version 3.3.4723. 1 (27-Sep-20 12); and Analyze build 290 (24-Sep-20 12).

Preparation of sample sets was performed by Mavis Date-Chong (Book nos. 495, 504, and Appendix pages 1-10 of these notes.)

DNA preparation • Liquid whole blood was obtained from a female contributor (profile QCAO 18). • Liquid semen was obtained from two male contributors (profiles QCA I 08 and QCA I 00). • DNA was obtained from these samples through a ProK/SDS digestion, phenol/chloroform/isoamyl alcohol

extraction, and Centricon YM-30 (Millipore) concentration into Tris-EDTA (TE") buffer. For the semen samples, OTT was added to the digestion buffer to aid in sperm cell lysis.

DNase I digestion • To examine the effect of degradation on each program's ability to detect a contributor, a degradation series

was prepared using DNase I (Invitrogen, Life Technologies) and the DNA extracted from male contributor QCAI08. Following an approach previously described (Swango et al. Forensic Sci. Int. 158 (2006) 14-26), a series of digests representing a range of degradation levels was selected from two sets of multiple time points and enzyme concentrations (no digest ion ; 1.7 x 10-4 Uing, digested for 6, 12,20,30, and 55 minutes; 3.3 x 10-4 Uing, digested for 6, 12, 30, and 55 minutes.)

o DNA concentrations were determined by a qPCR assay (Hudlow et aI. , Forensic Sci. Int.: Genetics 2 (2008) 108-125) that utilizes the THO I STR locus as a target sequence to assess total human DNA. The THOI amplicon size of approximately 170-190 bp is in the middle of the molecular weight range for most mUltiplexes. The assay also includes a 67 bp target sequence flanking the

Page I of25 epic.org EPIC-16-02-02-CalDOJ-FOIA-20160219-Validation-Study-Diff-Degradation 000001

STRmix V 1.0.7.49 TrueAliele Casework


December 10,20 14

CSF I PO STR locus. The ratio of the results from these two targets was one measure used to assess degradation when selecting the time points noted above.

o Because the qPCR THO I target is a mid-range STR locus, it is expected that the degraded contributor will present as the majority contributor at low molecular weight loci and as the minority contributor at high molecular weight loci. When the genotypes of the contributor are known, this would be observed as a decrease in the degraded DNA donor's mixture proportion (Gill et al. Forensic Sci. Int. 91 (1998) 41 -S3) as the molecular weight of the loci increases. The slope of the linear regression line for the Mx values across loci is another measure used to assess degradation in these test samples.

Mixture preparation • Two-person differential degradation mixtures (Book no. S04) were created as a I: I combination ofRI

(intact DNA) and R2 (intact or degraded DNA). • A three-person differential degradation mixture was created as a 6:3: I combination of, respectively, R3

(intact DNA), R2 (degraded DNA; 3.3 x 10" U/ng, digested for 6 minutes.), and RI (intact DNA).

PCR amplifications and capi llary electrophoresis • The two-person and three-person differential degradation samples were amplified using I ng of total

template DNA. Amplifications were done in duplicate using the AmpFISTR Identifi ler Plus multiplex and the Applied Biosystems GeneAmp PCR System 9700 thermal cycler using 9600 emulation mode under standard manufacturer conditions. Detection was by Applied Biosystems 3130XL Genetic Analyzer with S-second injections at 3 kV. Where appropriate, fsa files fragment sizing and allele calling were performed with GeneMapper ID v3.2.

• 2-person mixture Run folder

o RF _2 IMCE268_0SI71 UDPLUS_mixtr-deg Injections used for study (first injection for each amped sample); folder " [ ... ]\2PM DifDeg STRmix\2PM DifDeg STRmix Study Samples" :

o H04_LADDER_004_21 MCE268_0S171 UDPLUS_mixtr-deg.fsa • All ladders overlaid in GM: Sizing appeared uniform. • I selected a ladder with the high RFU results that appeared to have average sizing.

o C04_9947A_003_2 I MCE268_0S171 UDPLUS_mixtr-deg.fsa • Positive amplification control

o D04_NC_004_2IMCE268_0SI71 UDPLUS_mixtr-deg.fsa • Negative amplification control

o BO 1_2_ QCI08-S00pg_002_21 MCE268_0S171 UDPLUS_mixtr-deg.fsa • Male

o AOU _QCI8-S00pg_001 _2IMCE268_0SI71 UDPLUS_mixtr-deg.fsa • Female

o EOI _S_QCI8-QCI08jntact_001 _21 MCE268_0S I711 _IDPLUS_mixtr-deg.fsa o FO 1_6_QC I 8-QC 108_intact_002_2 I MCE268_0S171 UDPLUS_mixtr-deg.fsa o A02_8_QCI8-QCI08TubeS_OOI _21 MCE268_0S17 1 UDPLUS_mixtr-deg.fsa o GO 1_7_QC I 8-QC I 08TubeS_003_2 I MCE268_0S17 1 UDPLUS_mixtr-deg.fsa o B02_9 _QCI 8-QCI08Tube6_002_2 I MCE268_0S I71 1_IDPLUS_mixtr-deg.fsa o C02_ IO_QCI8-QC I08Tube6_003_21 MCE268_0S171 UDPLUS_mixtr-deg.fsa o D02_ II _QCI8-QCI08Tube7_004_21 MCE268_0SI71 UDPLUS_mixtr-deg.fsa o E02_ 12_QCI 8-QC I 08Tube7_00 1_2 I MCE268_0S 171 UDPLUS_mixtr-deg.fsa o F02_ 13_QCI8-QCI08Tube8_002_21 MCE268_0S171 UDPLUS_mixtr-deg.fsa o G02_ 14_QCI 8-QCI08Tube8_003_2 I MCE268_0S17 1 UDPLUS_mixtr-deg.fsa o A03_ 16_QCI8-QCI08TubelO_00 1_2IMCE268_0S I711 _IDPLUS_mixtr-deg.fsa o H02_ IS_QCI8-QCI08TubeIO_004_2IMCE268_0SI711 _IDPLUS_mixtr-deg.fsa o D03_ 19_QCI8-QC I08Tubel l_004_2 1 MCE268_0S171 UDPLUS_mixtr-deg.fsa o E03_20_QCI8-QCI08Tubell_OOI _2 1 MCE268_0SI71 UDPLUS_mixtr-deg.fsa o A04_ 23 _ QC 18-QC I 08Tube 12_001 _21 MCE268_0SI711 _IDPLUS_mixtr-deg.fsa o B04_24_QCI 8-QCI 08Tube 12_002_2 I MCE268_0S171 UDPLUS_mixtr-deg.fsa

Steven P. Mye~/b/~'1 ,"t L

of2S epic.org EPIC-16-02-02-CalDOJ-FOIA-20160219-Validation-Study-Diff-Degradation 000002

STRmix V1.0.7.49 TrueAllele Casework

Casework Interpretation of Comp lex DNA Mixtures PART I: Differential degradation studies

December 10, 20 14

o 803_ 17 _QCIB-QCIOBTubeI4_002_2 I MCE26B_05 I 71 UDPLUS_mixtr-deg.fsa o C03_ 1B_QC I B-QCI OBTubeI4_003_2 I MCE26B_05 171 UDPLUS_mixtr-deg.fsa o F03 _2 1_ QCI B-QC I OBTube I 6_002_2 I MCE26B_05 171 I _IDPLUS_mixtr-deg.fsa o G03_22_QCI B-QC 10BTube I 6_003_2 I MCE26B_05 171 UDPLUS_mixtr-deg.fsa

• 3-person mixture injections used for study (first injection for each amped sample): Run Folder

o RF _23MCEIB5_ 121913_CAFDAsmpls Injections used for study; folder "[ ... ]ISTRmixI3PM DifDegI3PM_DifDeg_STRmix"

o D02_LADDER_004.fsa • Allelic ladder

o AOI _QCIB_RI _OOI.2.fsa • "I" parts contributor, female, intact DNA

o CO I_QCIOB_R3_003.2.fsa • "3" parts contributor, female , degraded DNA

o DOI _QCIOO_004.2.fsa • "6" parts contributor, male, intact DNA

o HOI PC 004.2.fsa • Positive amplification control

o GOl _NC_003.2.fsa • Negative amplification control

o EO I_EM I_00 I.fsa o FOI _EMI_002.fsa

Evaluation of qPCR degradation ratio and Mx slope : 2-person differential degradation mixtures • Mx values for the 2-person differential degradation mixtures:

o The relevant electronic sample file s are in the folder "[ ... ]I2PM DifDeg S TRmixl2 PM _ Di fDelL Mx _Estimation"

o Calculated using spreadsheet "TubeXX.Mx Var (Dil Series).xlsm" (a copy was saved for each amplification.) This worksheet uses the true genotype combinations, alle le heights, and Gill et al. residuals calculation to establish, on a per-locus basis, the Mx that yields the lowest residual. Where the residual minimizes, this is the Mx for which the data best fits the genotypes in question. Note: No adjustments for overlapping stuner werej1.lade when entering allele heights.

o Locus values were ploned: ~r!Il\is is the locus Mx,j;;M1S is the average allele size in bases for that locus. (Note: These are ba?ea upon the true allele ?il!e;·not the perceived size affected by mobility modifiers.) The metric for further consideration is the slope of the inter- locus linear regression line. See spreadsheet "TubeXX.DifDeg Male Mx Graph.xls" (a copy was saved for each amplification.)

QCOIB + QCIOB Intact Amp I Amp2

~ '---f-o --o 0 0 0

o e 0 o 0

o o o Q

2] 0.6

~ • • , t-----------------------------------! .. ., ., t----., -

I

y • .o.OOOOIlSOllx .0.621314979 ft' . O.OOO3ISII3

, ~I--------------------------------~ • ~ _ _ m _ _ _ _ _ _ _ _

............ B ... $I •• 01 Oeteultd An-I"

Steven P. Myer~!I'1 ,-vI(/?

., •• . ,

2]0.6 ~ .. , ! .. ., . , .,

0 0 0 0

0

_ .

o 0

o o

y • 0.0001184190<.0.571557140 ft' . Q,055S02280

o

o o

• ~ _ _ m _ _ _ _ _ _ _ _

A .... ,.... a. .. Sli. 01 Detected AII.III

of25 epic.org EPIC-16-02-02-CalDOJ-FOIA-20160219-Validation-Study-Diff-Degradation 000003

STRmix V1.0.?.49 Casework Interpretation of Complex DNA Mixtures December 10,20 14 TrueAllele Casework PART I: Differential degradation studies

QCOl8 + QCI08 Tube05 Amp I Amp2

"' "' M ...

0 0

"' ~ "' o 0

0 00 -i 0.6 i 0.6

~ 0

0

10.5 0 0 0

j O.S 0

0 .~-

~ "' 0 0

~ ., 0 0

0

"' "' y • .o.00081999h +0. 14950057' y • .o.OO UX05O'Jll+ 0.793131967

"' II' .0,3935<'3790

"' R' - O.SlI2J3812ll

"' "' ,~~~~,~

" n' '" ". n, '" '" no '" n' '" '" '" " '" '" ". n' '" '" no '" n, '" '" '" AVI,..._ a. .. Siu 01 OtIt.ctMl An.r.. "".r., •• , .. Sill of O~t.ct.,;l AU.1et

QCOl8 + QC I08 Tube06 Amp I Amp2 , -

"' "' ... "' 0 ,., "' ':... 0

r-~ 0 0 i 0.6

0 i 0.6 0

0 0 1 0 j iU " "' ~

v

~ "' ! "' 0

I- 0 0 0 0

0

"' I---- "' y.-G.000983 19Ooc. O.73~

"' "' 11' . 0.519118090

y • .o.OOI I'lO&360. 0.792337447

"' R'.O,SB9926S25

"' , .. , , . ..... • TT' __ ~ ' _"'-r-'-

" n, '" '" n' '" '" no '" n, '" '" ". " n, '" '" n' '" '" n. ". '" '" '" '" Av,r.,. hM 51'1 of DllertH AlI,ln "WI'I,' .... 51'1 of OeIKIMI AlIeIet

QCO 18 + QC I 08 TubeO? Amp I Amp 2

" 0 "' "'

0 .. 0

0

"' 0 "' 0 0 0 0 0 0 0 :i 0.6 i M

j 0.5 0

0 1 0 0 " .. 0 0

~ "' 0 ~ "' 0 0

0 0

"' "' y . '(),OOt42062&. 0.'707)31«1

"' "' II' _0.&60849\56

y • .o.OOl60706h ' 0.929134595

" 11' . 0.614'74780

"' '~"r~~"-'-'Y . • . T, -~~'~T--r_

, .. '" .. -......--.- , .~

" n' '" '" '" '" n' n. '" '" '" '" '" " '" "' '" '" '" '" n. ". '" '" '" '" AVII'I" BaM SlI, of OeIKtMi Allel., """nt_ h .. SlI<I of Collided AIIeI ..

Steven P Myers~:~~ I~/' ('




December 10,2014

QCO 18 + QC I 08 Tube08 Ampl

., ·, t--coc---------------------------------------.' t---o-~O~~ 0

id6 t----------""-~~0~ ____ e-------~0------------------

j d5 t-----------------------7~~~-------------~ ., ~ ~ ., ., r----------------------,-.• ~.OO=c"~'"~'=h~ •• ~'~,,~"~,~~, ------., k' .0.111301422

, . • _ _ _ rn _ _ _ "" '" _ ill _

,,~ .... Soo .. SUI of DeIWItd AI .. I ..

QCO l 8 + QCI08 Tube lO Amp I

.,

'" 0

., 0 0

i 0.6 >-0-. 0

.~o j 0.5 ~ ., 0

~ 0 ., .,

y • ·O.OO1697S!.4 •• 0.89759234'

., R' 00.77U5744'

• '" - "" "" ,.

'" - "" '" - '" -"Wfale Base Slrl 01 OfloKted ..... IH

QCO l 8 + QCI08 Tubell Amp I

., 0 .,

., 0 0 0

i 0-6 0

j 0.5 0 ., ~ ., 0

0 0

"' ., y • .(I.OO I919U4Ic' O.988B3606 ., k'.O.7!;4/i61512

, • '" ,. ,~ '" "" '" ,. ,~ '" - '" -AVlra,. Ba .. Sill of O,tlet'" "11111'

Amp2

o ., t-------------------------------------------

., , _______ ~~. 0

~ O'6 T 0 o~ a

· ·, t-----------------------~~-<~----~o----~o-----~ 0.4 0 0

., y ·'O.OOI14~·O.90!0805674

., t-__________ ---"R.OO.601564)8 o ____________ ~

., o .. , •. ,-~ rr " ~. ~ .--.---- ... "

• _ _ _ rn _ _ _ "" '" _ ill _

Averql .... Sill of o.tected Allelfl

Amp2

"' ., ., __ 0

o~ 0 i 0.6

~o j 0.5

~ ., ~ 0

., 0

., y. -O,OOIUn43l( . Q.946lSI41S

., R' .0.'18&48366

, .. • '" - ,~ '"

,. '"" - "" "" - "" -Ave,.... hie Sill of Detected Ahln

Amp2

., 0 .,

0 ., 0 0 0 i 0.6

~ .. , 0

~ ., 0 0

0 0 ., .,

, • .o.oo190.261 •• 0 .9n931265

., 11;' oO.8S209Jl'M

, ... ..--. ..--- .. , . - ~ -------- . • '"' ,.

'" "" ,.

'"' ,. '" no - '" ,.

Ave, ... 11 .. 1 Sill of Dtot.ct..:l Al .. 1 ..


STRmix V 1.0.7 ,49 Casework Interpretation of Complex DNA Mixtures December 10, 201 4 TrueAllele Casework PART I:

QCOl8 + QCI08 Tube l2 Amp l

n: I 0

M

., 0 0

:i 0.6 =-00

I .. , --0

~ .. ., ., •• r ' .o.oD2Ul109Ic t t .OBI69989

1l' · O.,S9. 6)l14

, ~ n, "" n' ." '" '" no '" '" AVI,... a. .. 51ze of Detlcted AII.I~II

QCOl8 + QC I08 Tube l4 Amp I

.. 0 ..

0 ., :i 0.6

0 0

~ ."' 0

• 0 0 0

! 0.4 0

0

I (Ut--

"' r • -0.002396(1)9:0' O.91J' IIIOlO .. 1I.'.O.tlU.aUi9

, ,. ,. ,. ""--'--'~-~ ~ '" '" n, '" .~ "" '" '" no

AV''',1 a. .. Sill of IhIte«" ",,-I ..

QCOl8 + QC I08 Tube l6 Amp I

.. 0

•• ., 0"

1 0.6 0 0

~ 0 0 ." 0

~ .. 0

0 .,

"' l y . -tI.OO2U312&ct l .08l2H990 II' . O.8S9S4~71. ,.

, -~ ~ n, "" n, ." .~ '" '" "" A ........ at .. Si .. of Oettctlld AIIto'11

Steven P. Mye~fl \of A It'

I~v'

'"

Differential degradation studies

Amp 2

•• 0

•• ., ~

0 oo~ i 0.6

~ j 0.5 0

~ 0 ~ .. ., o 0

0 ., y • .o.Q02 12161lK +I.01002 1!181O

1l· . O.U21~l!iO •• ~,,---.,

'" '" no ~ '" '" n, '" .~ n, no '" '" "" '" no A ........ at .. $Irl 01 Otol.cted Al .. ',.

Amp2

.. .. "'

0 0

:i 0.6 0

~ ."' 0 0

~ .. 0 0 0 0

0 0

., 0 0 "' y • .o.oo12x ' O.96 IS 0 .. II' _O.SOll

, .. '" ." no ~ ." '" '" ." .~ '" ,ro ~ n, '" '" '" Av ..... al .. 51'1 01 Dellem AI .....

Amp 2

.. 0 .. ...........

"' 00 ~ i 1).6

"'-...... 0

10.5 0 0

I .. o~ 0 .. y. -o,OO2~19011 ' \.0<42003861

0 II' .0 .• 20263515

0 "' 0 0 ..

, " ,." ,- ._;-. ,....-.-~ •• -, r ____ '--r-, __ r r-__

'" '" no ~ n, .ro '" '" ... '" '" '" no '" '" no Aft .... 'I .. SUI of Oe1«ted Anelll

of 25 epic.org EPIC-16-02-02-CalDOJ-FOIA-20160219-Validation-Study-Diff-Degradation 000006

STRmix VI.0.7 .49 TrueAlIele Casework


December 10, 20 14

• As demonstrated in the electropherogram for sample QCOl 8 + QC I08 Tube08 where the Mx slope for the degraded contributor is -0.00 17, the shi ft in the Mx may not be readily discern ible when simply looking at the mixture. It' s only when the true profil es are known that the di fferential degradation becomes apparent.

......... OC-QCI._JOOt.JUlGMO ..... m..1l,. 1_ M"'!! !PIII! 171 - i ICI'@ ,. ,.

~~.I .. - - • M ~ • • • - M • - -

:~I ~~~ ~ ~ ~ ~~ I

.... .J_qnuolllN..MUIJftZHI ..... m ..ll ... ..,J~ II!!!I II!! III!!! I!!! IZ!- I !CIII!!!:!

'~I '" '. ,. ,. • M m M • · • m - - • .. - •

~ QCO 18 (female, intact)

I & ~~ M ~ ~

1Il.J_~..-.u.llft."·I. I-IIIU!! 1111111 I!I I!!11 I !CIIII!!!

'r '. - '. M .. - - - M • • • - .. • • - -

·1 . ~ Q'"'' (m.'" ''''''d)J!

~ ~ ·u I

...... ·UItu-q"· _ _ .. ..1·lI._ ... "'"..D • -III,!'. ,1M' i 1!!!l@Ii iWP' I g il'" I ", '. ,. - '" - .. ,. ,. - ", m · -

~r .. ..

_ .. ,. m . - m

'~I ~ ~ ~ M ~~ ~ ~ ~ MV_~..-u.Jll'l.".I . -mill!!! rm!!1 1111111' 10- I IIIIIII! - '. '. ,. M .. - ,. - - • .. M • • • - M - .. .. M . • - • - ,.

:~ .~ ~ M ~~ ~ ~ _.l.Q~ ..... ftUPI."' l. I-

1111111 11111 I IIIIIII! IIIIIIII! I IIIllm I - '" ,. '. ,. .. ,. ,. ,. - - ". m • .. · - M - • .. .. .. .. - .. :~I

~1 ~~ ~~ 1 ~ ~ ".

Steven P. Myers ~l>{ of 25 epic.org EPIC-16-02-02-CalDOJ-FOIA-20160219-Validation-Study-Diff-Degradation 000007

STRmix V 1.0.7.49 TrueAllele Casework

.. ...,_Q .... QQ._ •• ~.OJ1mJlI •

A'!!H!I .. 0

~-t o~1 • o.

"'k~_..m.Jnt.".II. I_ P!f@J o. o. o. o. ~

O~I ~o~

I ..... o.

J .....

o.

1II.J_I;I~.JUIQKI_"nJJI","'.~

Casework Interpretation of Complex DNA Mixtures PART I : Differential degradation studies

, tItUS !AIl'" o. o. o. "' ~ ~ ~ o. * m

~ ~ o ~ M ~ i ,...,1 'il'''' o. O. M o. m

m _ - M - - m - -

~ & ~ .~ mpH" i Wi i tI.... i 1'''''

December 100 2014

• m - - -~

- m - - - - -

!f4 '

M1.J_~_"'""..DR-n"'l. I_ III!Id p." , ...

lIIU..Q-"'''''''''''''Il)II",n·l. I_ IiKI pig '"

~- ~-~ ---·1 •• m 1M

o. _ ., .

T


STRmix V 1.0.7.49 TrueAlle le Casework


December 10,20 14

• In order to evaluate the STRm ix and TrueAllele 2-person mixture results as a function of degradation, the QC I 08 qPCR degradation ratio and QC I 08 Mx slopes were plotted for each combination of DNase I units and time point.

•

Samgle Name DNase I Digestion gPCR - Sloge' uL units / uL units min deg ratio average

QC I 08 Tube05 4 0.2 0.8 6 2 9. IE-04 QC I 08 Tube06 4 0.2 0.8 12 2.3 1.1 E-03 QC I 08 Tube07 4 0.2 0.8 20 3.4 1.5E-03 QC I 08 Tube08 4 0.2 0.8 30 4.2 1.7E-03 QC I08 Tube lO 4 0.2 0.8 55 4.5 1.8E-03 QC I08 Tubell 4 0.4 1.6 6 5.6 1.9E-03 QCI08 Tubel2 4 0.4 1.6 12 7.6 2.1 E-03 QCI 08 Tubel4 4 0.4 1.6 30 5.3 2.3E-03 QCI08 Tubel6 4 0.4 1.6 55 6.7 2.6E-03

'Average for the two amplifications.

Plotting the 2-person differential degradation data 0 The two y-axes were scaled to promote graphical overlap. 0 Linear regression lines: Solid for qPCR degradation ratios, dashed for Mx slopes.

8

7

6

0 .~ 5 a: c 0 'il4 " ~ k3

2

1

0

•

• A ··· ..., ... ..

• O.oo23x + 6.2397 Rl =0.0024

.o·Er

•

~

~'+1.988S ~ R' .0.824

fv -~ "iE :05~ +o.Oiii9] L. ___ ~~ _~ _~:~?_~; ___ .J

... ..

3.E-3

2.E-3 ~

~ ~

g 2.E-3 :

~ ~

~ ~

1.E-3 0 ~ g. ~

8.E-4

-l----_---_---_---_---_-----I- 3.E-4

0 10 20 30 40 50 60

Minutes of Digestion

. 0.8 units - qPCR . 1.6 units - qPCR 0 0.8 units - Mx 6 1.6 units · Mx

• For the purposes of th is study, the slope of the Mx increased consistently as a function oftime and/or DNase I quantities. The qPCR degradation ratio and Mx slope values gave almost identical results for 0.8 units of DNase 1. The regression lines overlap and have similar R'- However, the Mx slope had a much higher correlation to increases in time for 1.6 units. Additionally, the slopes of the 0.8 and 1.6 linear regression lines for the Mx slope values are similar, unlike those of the qPCR degradation ratios.

• Graphing of the STRmix and TrueAllele results will use the Mx slopes for the y-axis.

Steven P. Myers ~1(//ltJ of25 epic.org EPIC-16-02-02-CalDOJ-FOIA-20160219-Validation-Study-Diff-Degradation 000009

STRmix V 1.0.7.49 TrueAlIele Casework


December 10,20 14

Evaluation of Mx slope: 3-person differential degradation mixture • Mx values for the 3-person differential degradation mixture:

o Calculated using spreadsheet "3 PM _ DifDeg Mx spm 01.24.20 14.xls" (a copy was saved for each amplification.) This worksheet uses the true genotype combinations, allele heights, and Gill et al. residuals calculation to establish, on a per-locus basis, the Mx that yields the lowest residual. A 3-dimensional map is created with the Mx for contributor I on the x-axis, the Mx for contributor 2 on the y-axis, and the residual on the z-axis. (No graphing is needed for contributor 3, since that Mx becomes fixed once you know the Mx values for contributors I and 2.) The bottom of the cavity (or trough for a map where two of the contributors have the same genotype) is the point where the residual has minimized. These are the Mx values for which the data best fits the three genotypes in question. Note: No adjustments for overlapping stutter were made when entering allele heights.

o Examples of residuals maps:

All genotypes different Two genotypes the same, one different

"

i I

'lbl • ..' ""'1

o Locus Mx values were plotted for each contributor: 'k:.a;.ds is the locus MX,.~:;,,!>iS is the average allele size in bases for that locus. (Note: These are tra's<l'd upon the true aller~i~e, not the perceived size affected by mobility modifiers.) The metric for further consideration is the slope of the inter-locus linear regression line. See spreadsheet "3PM_ DifDeg Mx Summaries spm 01.24.2014.xls" .

Amp I Amp2

0.8 ,

.,

'"

•

y. S.70[.()4 .. S.10E·Ol fI' . I .S9E~1 .

•

A Y ' 6,50( -04.- 9. 1~ E -Ol D

fI ' - 4.15[ ·01

ISO 200 no 300

Allera,e (True) BlIISl! Size of Genotype', Alleles

OMM(6PM SI OMM !JINR .. d.trocMdJ 6 ~""'MII INRJ

••

.,

., • '"

• • • ---<------ . v _ 5.l7(.4)oI"S.16£"()1

R' . 1.43[..()1 .

• • • o 1::1 ' · · Iffi[-Oh ....... [ .() !

~~~o~~~::o:=~:=~:O=='~.;. ';:~~:'~~~ ,, 0 0

0

&7= ~~ 1.1> 1. Ji 0 IJ

V . 1.76£_0.\ • • 6.98[.(12 t.

• fl. ' _ 1.34[ '()1

ISO 200 2SO !OO

Alle""e (True)8ne Sire of Genotype's Alleles

• Graphing of the STRmix and TrueAlIele results will use the Mx slopes for the y-axis.

0f25 epic.org EPIC-16-02-02-CalDOJ-FOIA-20160219-Validation-Study-Diff-Degradation 000010



Creation of mixture interpretation input files

December 10,2014

• STRmix - Import fil es were created as GeneMapper ID exported Genotype tables (txt format) per the STRmix recommendations and settings.

o Reference import fil es included the following information: sample name; marker name; and allele designations and the measured size in bases for the genotype's allele peaks.

o Mixture import files included the following information: sample name; marker name; and allele des ignations, measured size, and heights (RfU) for all possible allelic and reverse stutter peaks.

o 50 RfU analytical threshold. o Peaks consistent with artifacts other than reverse stutter were electronically deleted prior to the

table export. o Exported tables did not always maintain the proper locus order. To correct this, the txt files were

reordered using the Excel spreadsheet "LocusOrderForExportTXT.xlsm". o Evidence import file : 2PM_DifDeg_STRmix.txt o Reference import file: 2PM_DefDeg_STRmix_R.txt

• TrueAliele Casework - This system incorporates its own data analysis component that uses the original fsa files for the samples of interest, allelic ladders, and amplification controls. Individual capillary's files were combined into a virtual gel for sizing and allele calls. The resulting information was then uploaded to the server.

o Mixture settings: • "newdd _acquire" set to File mode. • Identifi ler template • AB13130xl • Ladder assignment: closest • Size standard

• 2-person: GS500 (MDC didn 't run this using the final lab protocol) • 3-person: L1Z600

• Controls: Names (Ladder, 9947 A, NC) o Analyze Module

• All size standard peaks labeled. • No rules fired. • Gels:

• 2-person: TACM DCDifDegStudyData.gel • 3-person: 3PM_DifDeLRf.gel

o Data Module Uploaded gel to world: Rosifume

Interpretations • STRmix and TrueAliele Casework interpretations were performed in triplicate. Both amplifications were

interpreted separately and as a joint interpretation. In some, but not all , instances where likelihood ratios reported by TrueAliele Casework were observed to be identical to the ninth decimal place, an additional interpretation was performed, and one of the duplicated results was dropped from furth er consideration . Such instances were not limited to major contributors with 100% probability attached to all of their genotypes.

• In general, both systems use similar approaches to mixture interpretation and the calculation of the likelihood ratio. However, they differ in myriad details. Rather than try to limit the differences, the systems were compared "as is". They were, however, interpreted with more MCMC cycles than standard.

• Both systems were interpreted with no assumed contributors. Initial interpretations were based solely upon the mixtures.

• STRmix interpretation: o The 2-person differential degradation study was tested twice . Once using the allele variance

settings calculated by Cal DO! using Model Maker vers ion 1.0.7 .49, and a second time using the allele variance settings supplied by ESR, calculated using the same Cal DO! data and the Model Maker version found in STRmix V2 .0.

Page II of25 epic.org EPIC-16-02-02-CalDOJ-FOIA-20160219-Validation-Study-Diff-Degradation 000011



December 10,20 14

a The 3-person differential degradation sample was only tested using the ESR settings. a Settings

• Variance • Model Maker VI.0 .7.49: 1.62

For interpretations using this variance, see fo lder "[ ... ]I2PM DifDeg STRmixl lnterp"

• ESR supplied using approach in Model Maker V2.0 : 3.392 For interpretations using th is variance, see folder "[ ... ]I2PM DifDeg STRmix\interp New Variance"

• Note: V2. 0 Model Maker approximately doubled the allele variance. o Detection threshold: 50 • Stutter: 0.3 o

o

o

o

o

o

o

o

o

o

o

o

o

o

o

o

o

o

Degradation: 0.02 Drop-in: 0.0 Drop-in parameters: 0.0,0.0 Clip rate: 0.0 I Extreme cl ip: 0.005 Saturation: 7500 Use ref a lleles: N Use locus amp factors: Y # MCMC chains: 4 Extended output: N MCMC accepts: 100,000 Bumin accepts: 20,000 HPD iterations: 1,000 Sig value: 99.0 Sides: I Alleles per locus: 10 Locus amp variance: 0.022 Default kit: lOP _3130XL

• TrueAllele Casework interpretations. a Request Module

o Settings: • Process: twounknown • Part: evidence • Defer: no (see the note below reo the 3-person mixtures) • Bumin: 100,000 • Readout: 100,000 • Offiadder: short • Degraded: on • Logging: off • Sort: off • Overwrite: no

o 2-person mixture requests • Request: TAC_MDC_DifDeg2a. req

o Joint interpretation of two amplifications. a Duplicate amplifications joined at the item level, so that they will be

based upon one mixture weight. • Request: TAC_MDC_DifDeg2a. l.req

a Additional requests. a In most cases, this was just a third request. For Tube 14. 17. 18, a fourth

request was performed, because the first two gave identical LRs (they might have randomized from identical seeds.)

• Request: TAC_MDC_DifDeg3 .req

Steven P Myers ~~ It '-I of25 epic.org EPIC-16-02-02-CalDOJ-FOIA-20160219-Validation-Study-Diff-Degradation 000012

STRmix VI.O.7.49 TrueAllele Casework


December 10, 20 14

o The interpretation of each amplification was performed separately. • Request: TAC_M DC_DifDeg3.l.req

o Additional requests were performed. In most cases, this was just a third request. For Tube 12.23, a fourth request was performed, because the first two gave identical LRs (they might have randomized from identical seeds.).

• Request: TAC_MDC_DifDeg3.2.req o Additional requests were performed for the second amp. o All samples were cycled through prior to starting the second set of

request. This was done to avoid duplicate requests getting the same seed.

• 3-person mixture requests • Request: TAC_3PM_DifDeg.req

o Separate and joint interpretations of the two amplifications. o NOTE: The Defer setting was "yes" for the duplicate mixture requests.

After the requests were initiated, and the first set of mixture requests were running, the second set of requests was activated.

• Request: TAC_3PM_DifDeg.l.req o Additional requests were performed.

o Report Module ~,

• 2-person mixtures: No overly narrow mixture weightl distributions were observed during a review of all interpretations. Some interpretations had the chains switch weights, leading to distributions with large amounts of overlap. Other interpretations had little to no overlap observed.

• 3-person mixture: With one exception, the interpretations had chains that appear to be reasonable, with two of the contributors swapping higher-lower order. The 3rdjoint ~!2te ,;e:ta~on had rope-like chains until the end of the run:

___________ \11,"' ... "1,

________ irJ..tv --_ ._- ... .... ..;.,. -

• Detailed reports were exported as Excel files for import into DOl LR spreadsheets (see below).

Steven P. Myers ~, C ,1/( t-[ j""t- I




December 10,2014

Summary: Number of interpretations per mixture and interpretation system

1st Amp 2nd Amp Joint Interpretations Interpretations Interpretations

2-Person Mixtures STRmix TAC STRmix TAC STRmix TAC • • •

QC IS-QC I OS intact 3 3 3 3 3 3 QC IS-QC I OSTube5 3 3 3 3 3 3 QC IS-QC I OSTube6 3 3 3 3 3 3 QC IS-QC IOSTube7 3 3 3 3 3 3 QC IS-QC I OSTubeS 3 3 3 3 3 3

QC IS-QC I OSTube I 0 3 3 3 3 3 3 QCIS-QCIOSTubell 3 3 3 3 3 3 QC IS-QC I OS Tube 12 3 4 3 3 3 3 QC IS-QC I OSTube 14 3 3 3 3 3 4 QC IS-QC I OSTube 16 3 3 3 3 3 3

3-Person Mixture STRmix TAC STRmix TAC STRmix TAC EMI 3 3 3 3 3 3

• All STRmix interpretations performed twice under separate allele variances.

Likelihood Ratios • Likelihood ratios were calculated at e ~ 0.0 I using FBI African American, Caucasian, and Hispanic

databases (JFS 199944(6); FSC 1999 1(2); FSC 200 1 3(3); each adapted as necessary for use in the specific software package, for example by removing U<" and ">" binned alleles.)

• Electronic reports were generated for each system: o STRmix "[ ... LResults.txt" files that include genotype combinations and associated donor

combination weights, but no LRs. o TrueAllele "detailed reports" (xIs files), with all values to 9 decimal place, for each interpreted

contributor to the mixture. The confidence level was set to 1.0 so that 100% of the interpreted genotypes (those assigned any probability by TrueAllele) would be included in the report. Note: In the initial part of the study, separate reports were created for each combination of reference-con tributor-population database based upon a manually selected reference-contributor pairing. Later, as allowed by a change in the LR calculation spreadsheet, only a single referencepopulation was necessary for each contributor. These latter reports were all imported into a single spreadsheet for automated contributor selection by the program (see the discussion below.)

• Electronic reports were imported into an Excel spreadsheet ["STRmix Val. Non-Contrib. (spm 1.IS.14).xlt", "TAC Val. Non-Contrib. (spm 4.9.20 13).xlt", and "TAC Val. Non-Contrib. (spm I. I S.20 14).xlt"]. The January IS, 2014 versions also required the import of a csv file containing a known contributor's profile. Each spreadsheet applies the approach of its respective system to calculate the LR.

o The TrueAllele reports imported into spreadsheet "TAC Val. Non-Contrib. (spm 4.9.20 13).xlt" were based upon a manual selection of the interpreted contributor for each comparison (i.e., whether QCO IS was assigned contributor I and QC I OS was assigned contributor 2, or vice versa, was based upon a manual review of which order gave higher LRs for these two references.)

o For spreadsheets "STRmix Val. Non-Contrib. (spm 1.IS.14).xlt" and "TAC Val. Non-Contrib. (spm I.IS.20 14).xlt", the assignment of a particular interpreted contributor to a comparison reference was done automatically based upon a single LR that is a combination of the three population LRs:

Steven P. Myers ~~~ 1"\ rz,-I t j (


STRmix VI.0.7.49 TrueAllele Casework


December 10, 20 14

• The STRmix spreadsheet uses a LR created from the sums of the three popUlations numerator and denominator conditional probabilities

NumAfAm + Numcauc + NumHisp LR s = --;:--"='---c-::;---'=--:-:::----"=

DenA! Am + Deneauc + DenHisp

This value is also calculated by STRmix as a stratified LR (in this case, all populations would be in equal proportion.) It is noted that STRmix assigns the highest populationspecific LR to each comparison regardless of whether or not all population LRs will represent the same interpreted contributor. Therefore, at times the values used for the study at hand may be sl ightly lower than those observed using the LR calculating function in STRmix. This difference should be limited in scope, since a given reference should only have similar LRs for multiple interpreted contributors when the contributors are present in similar proportions as in , for example, a I: 1 mixture.

• The TrueAllele spreadsheet uses the harmonic mean of the three population LRs:

The harmonic mean is not part of the TrueAllele approach. TrueAllele requires the user to assign a specific reference to a specific interpreted contributor. For some mixtures, this can be a simple matter [e.g., when each reference has LR» I for one interpreted contributor and LR« I for the other interpreted contributor(s)]. For other mixtures, multiple contributors may give their maximum LR for the same interpreted donor, or a reference may give their maximum population-specific LR in different contributor (as also happens in STRmix; see above.) In TrueAllele, unlike in STRmix, there is no ability to calculate a single LR for multiple references jointly.

,II'! Steven P. Myers ~ Ie I' '-\ of25 epic.org EPIC-16-02-02-CalDOJ-FOIA-20160219-Validation-Study-Diff-Degradation 000015



December 10,2014

SENSITIVITY: Graphing the 2-person mixtures • Populat ion-specific LRs for each comparison were imported into the following spreadsheets

o STRmix 2PM~DifDeg Graphs (spm 06.24.2014).xlsm o TAC 2PM~DifDeg 0 and 0.01 Graphs (spm 06.24.2014).xlsm

• Harmonic means (LRH described above) were calculated for each comparison. • The results were plotted as the Mx slope on the x-axis and the log,o LRH on the y-axis.

o Note: Since you can ' t calculate a log ofLR ~ 0, those values were assigned the value of~5 .

• QCOl8 (female, intact DNA) ~l>"'"

STRmix (V 1.0.7.49 alle le variance) (. (()Z-O eolhAmps C AmpI Il. Amp2

A 0

. 1.(-04

o 0 • 0 D O 0 0

Slope of Locus Mil for DeSnlded Slimple

., .

· 10 -

For '''phi ..... 'Both Amps" u .... naverql of 1M "Amp I" ,nd "Amp 2" MDpu.

TrueAllele (No locus LR thteshold)

~ .

,. .0 0 • - ' 5, 0 " .

~ ! § o · 8 n o o § o. • W ·

.~ B o·

5 ~ . 0 0 0 • , .

f 8 0 • 0 •

~

1'1 • . 1.(.(1) lJ. 0----.

-!t.(J3 .).(.(13 .2.(-03 'IJ~ -1.(.04 • 2.(-04 " .. ..... '0 cc:a<I4J O O .6. 0 0 ., .

-\0 -slope of locus Mil for Oe,raded Slimple

for ,,,phln .. ' 80th Amps' u ... , n """ .Blol 1M "Amp I" ,1\11 "Amp 2" sIopH.

<3(n"'1 STRmix (V2.0 allele variance) ~.2,9"Z--

O loth""'," CAmpI OAmpl

slope of locus Mil for Delrllded Sample

,. ' 0 o .

" " , .

., .

· \0 •

for .",*,,,,, "8o!hAmpS' uHS In IY"~' of the ' Amp I" .nd "Amp 2" sID,,".

TrueAllele (Locus LR threshold = 0.01) 0 80thAmps O Amp I l). Ampl

~ .

,. .0 0

• ~ g ~ 8 " . q § 0 · 8 n 0 0 § o.

8 0 .6.0 " .

:~ ~

i~ t o g& f 8 0 0 0 o· 1'i • ·LE.03 II t.0l • -!E.(13 ·l.f-m -u-m ·1.f.()4 .l.(.(N

0 s ., .

·10 -Slope oflO(u. Mil for Oelraded sample

For .,.phing. "80th Amps" UMSln _",Iof lhe "Amp ' " and "Amp 2" oIopes.

" ..

Steven P. Myers~/1 "I of25 epic.org EPIC-16-02-02-CalDOJ-FOIA-20160219-Validation-Study-Diff-Degradation 000016



December 10, 2014

• QCO 18 (male, degradation series)

STRmix (V 1.0.7.49 allele variance) STRmix (V2.0 allele variance)

• ~

• J ~ l i! ~ ;

O &oth.t.mps D Amp l !lAmp]

" . ;, 0

]i .,~--__ ----______________________ ~r-__

r+ 0 0 o 0 l!.. t:. 0 0 0 0 0 . , .

Slope of l ocus Mx for DeIAi ded SIImpl' ·10 -

For ,tt l/hi"" ' 80~h Amtn" !,1M' , n I"" "tof the ' Amp I" .nd "Amp l" . IoPH.

" .. ., .

-10 -Slope of Locus M. for Oql'1lded Simple

For'f'phi"" ' 8othAm",' UltUn ........ . oftht "Amp 1". nd ' AmpZ" lIopu

" ..

TrueAllele (No locus LR threshold) TrueAllele (Locus LR threshold = 0.0 I)

Slope of Locus Mx for De, raded SIImple Slope of i.«us Mlt for Oe,.,.d,d SlImpl, ·10 •

for .",,1>1 ..... "80th Amps' !,IMS .n.",,,s" 01 tho "Amp 1"M "Amp 2" oIopH. f Ol lllphlna. "80th Amps' ... H1ln _"1' of tile "Amp I" .nd "Amp Z" sIoI>n.

• As a point of comparison to our current approach, values using MixMaster were plotted in the same way. o Only one interpretation was performed for each comparison.

• MixMaster has a randomization seed, so the results would be identical if started fresh from opening the spreadsheet.

o Interpretation performed using "MixMaster \DP (Re\. 1.1 ).xlt" • No joint interpretations of the two amps were possible. • Note: When MixMaster was run using the defaul t Mx calculated from 4-allele loci,

Tubel6 amp I had a Type I error (false exclusion) at D2S 1338 for QC108 . The Tube l6 amp I data included in the plots is based upon a reanalys is using the 3- and 4-allele Mx estimated using " Mx CALculator (Re\. 1.0).xlt" . With that Mx, QC I08 was properly included. All other amplifications ' interpretations are based upon the default Mx.

Steven P Myer~~ I' 4 of25

epic.org EPIC-16-02-02-CalDOJ-FOIA-20160219-Validation-Study-Diff-Degradation 000017

STRmix VI.0.7.49 TrueAllele Casework

Casework Interpretation of Comp lex DNA Mixtures PART I: Differential degradation studies

December 10, 2014

o The major and minor contributor allowed genotypes were compared to the profiles ofQCO 18 (female, intact) and QCI08 (male, degradation series). This comparison was done by examining for simple inclusion/exclusion, as well as by using the average Mx to see which contributor would fit them best.

o o o

o o

Tube

Intact Tube05 Tube06 TubeD7 Tube08 Tube 10 Tube I I Tube 12 Tube 14 Tube l6 Tube l6*

Amp I Female Male

Incl/~~ve Mx both minor

Incll~ Ave Mx both major

both minor both major both major both minor both minor both major both major both minor both major both minor both minor both major both minor both major

major major minor minor major minor no major maior minor minor maior

Amp2 Female Male

I n cl~Of'i'I AveMx both '"?~ minor

Incl,fE,rot ~ve Mx both~ major

both minor both major both major both minor both minor both major both major both minor both minor both major both minor both major both minor both major

major major minor minor major minor minor major

•• I~r H / " :C I A.>c:..~'i: . .D 1.-'Utb, "'The interpretation using the override Mx from Mx Calculator. ~t:';>'71-f ~~ ~:ro ,",~~~L~S. ,

LM.f 'VOci ~C~-r'UO c.::... ............... .rw1 The major and minor donor RMPs were calculated in "STRstatlD_DOJ_v0426 l2 .xlt". "'7"" LRs were calculated as IIRM P. Population-specific LRs for each comparison were entered into the following spreadsheet

• "MixMaster 2PM_DifDeg Graphs (spm 06.l2.20l4).xlsm" Harmonic means (LRH described above) were calculated for each comparison. The results were plotted as the Mx slope on the x-axis and the log,o LRH on the y-axis.

• For Tube 16, the LRH for both interpretations (standard Mx and Mx Calculator override Mx) were both plotted.

• NOTE: Tbe stippled minor donor data point in tbe Minor donor grapb would be LR=O for QC108, per tbe discussion above.

MixMaster - Major donor MixMaster - Minor donor

B

·1.[·03

o •

DAmpl o. Amp2

~ -

~ .

" . . ~ oft ,f'o • 0 ~ 0 mO .

, -

.2.[.()3

., .

Slope of lows Mil for De,J'lded sample

LRH Total (degraded mixtures only) LRH = 0 (standard Mx) LRH = 0 (Tube 16 Amp I alt. Mx)

Steven P. Myers ~\ ~Ill-f t"Z- l

~

1 i 'J

•

2PM 36

2.78% 0.00%

D,t.mp l o. Ampl

Slope of Locus Mil for De&J'lded sample

" . ~ ,

" . m

O ·

, .

· 10 •



Casework Interpretation of Complex DNA Mixtures PART I: Differentia l degradation studies

SENSITIV ITY: Graphing the 3-person mixture

December 10, 2014

• Population-specific LRs for each comparison were imported into the fo llowing spreadsheets o STRmix 3PM_DifDeg Graphs (spm 06.24.20 14).xlsm o TAC 3PM_DifDeg 0 and 0.01 Graphs (spm 06.24.2014).xlsm

• Harmonic means (LRH described above) were calculated for each comparison. • The results were plotted as the Mx slope on the x-axis and the IOg'0 LRH on the y-axis.

o The results for each contributor are encircled. o LRH ~ 0 was plotted as - 5.

• STRmix o The one LR~ 0 was for QC I08 (degraded DNA.)

o Both Amps D Amp 1 tJ. Amp 2

lS

20

® 15

- --------, , I '" 0 c ,

10 , I , - -----_ ... ... ,

s 11 C'~1 !~--~--~----~---r---,----~--0

~. ,(:,~.:~:~ '.038E-04 ·G.'·M ·4.'·MlEM .:'100 H-04 4.E-04 6.E-04 8.E-04

· 10 Slope of Contributor' s loe us Mx Values

For graphing, M60th Amps" uses an average of the ~Amp I " and "Amp r slopes.

• TrueAliele (Locus LR threshold ~ 0.01)

~

" g

o Both Amps DAmpl 6 Amp2

lS

20

15

10 ... -----!

'I Cl) · (: I ! 8) ~~--__ --__ --__ --__ --__ --__ --04---__ ~~~~~-'

'!f.03 · 1.E·03 · l.E·03 -8.[ -04 ·6.E·04 ·4.E·04 ·2.E·04 O.E+oo 2.E-04 4.E-04 6.E·04 8.E·04

· 5

· 10 Slope of Contributor's Locus Mx Values

Forgraphing, "Both Amps· uses an average of the "Amp I " and -Ampr slopes.


STRmix V 1.0.7.49 TrueA liele Casework


• TrueAliele (No locus LR threshold) o Both Amps DAmp 1 l:r. Amp 2

2S

20

15

December 10,20 14

-1.[-03 -1.E·03 ·aE·04 -6.(-04 .4.E.Q4 ·2.[ ·04 o.EtOO 2.[-04 4.[-04 6.E-04 8.E-04

·s

-10 Slope of Contributor's locus Mx Values

For graphing, "60th AmpsM uses an averageof the "Amp 1" and "Amp 2" slopes.

SENSITIVITY: Summary Tables • Population-specific LRs for each comparison were imported into the fo llowing spreadsheets

o STRmix _ Graphs_Data (spm 07.10.20 14).xlsm o TAC_Graphs_Data (spm 07.10.2014).xlsm

• Notes: o These tables do not include the results from the 2-person mixture differential degradations study's

"QC 18-QC I 08 _intact" mixture. o STRmix results using the V 1.0.7.49 variance were not included.

• All contributors for 3 replicates each of amp I, amp 2, and the joint interpretation:

Total LRH ~ O

0 < LRH < I LRH < I

STRmix (V2.0 variance) 2PM 3PM Comb 162 27 189

0.00% 3.70% 0.53% 0.00% 0.00% 0.00% 0.00% 3.70% 0.53%

TAC (0.01 minimum) 2PM 3PM Comb 162 27 189

0.00% 0.00% 0.00% 17.90% 0.00% 15.34% 17.90% 0.00% 15.34%

• All contributors for 3 replicates each of amp I and amp 2:

Total LRH ~ 0 O< LRH < I LRH < I

STRmix (V2 .0 variance) 2PM 3PM Comb 108 18 126

0.00% 5.56% 0.79% 0.00% 0.00% 0.00% 0.00% 5.56% 0.79%

Steven P Myers~c I' "/


0.00% 0.00% 0.00% 14.8 1% 0.00% 12.70% 14.81% 0.00% 12.70%

TAC (0 min imum) 2PM 3PM Comb 162 27 189

56. 17% 55.56% 5608% 4.32% 0.00% 3.70% 60.49% 55.56% 59.79%

TAC (0 minimum) 2PM 3PM Comb 108 18 126

49.07% 33.33% 46.83% 6.48% 0.00% 5.56%

55.56% 33.33% 52.38%




December 10,2014

o All contributors for 3 replicates each of the joint interpretations:

STRmix (V2.0 variance) 2PM 3PM Comb

TAC (0.0 I minimum) TAC (0 minimum) 2PM 3PM Comb 2PM 3PM Comb

Total LRH = 0 O< LRH < I LRH < I

54 0.00% 0.00% 0.00%

9 63 0.00% 0.00% 0.00% 0.00% 0.00% 0.00%

PRECISION: Graphing the 2-person mixtures o See the following spreadsheets

54 9 0.00% 0.00%

24.07% 0.00% 24.07% 0.00%

o STRmix 2PM_DifDeg Graphs (spm 06.24.2014).xlsm

63 0.00%

20.63% 20.63%

o TAC 2PM_DifDeg 0 and 0.01 Graphs (spm 06.24.2014).xlsm • LRH within like interpretations were compared in a pairwise manner.

54 9 63 70.37% 100.00% 74.60% 0.00% 0.00% 0.00% 70.37% 10000% 74 .60%

o E.g., the results of the three joint interpretations for a mixture were compared to each other. o The results were plotted as the lower of the two log,o LRH on the x-axis and the absolute value of the

difference on the y-axis. o The dashed line represents I log unit, i.e. , a factor of 10 difference. o LRH = 0 was plotted as - 5.

o QCOIS (female, intact DNA)

STRmix (V 1.0.7.49 allele variance)

•

t "

STRmix (V2.0 allele variance) OlolhAmps D Amp I OAmp2

"

__ -lI.m;x=: m ", m··iR~"~ ;o;a==··· .m··-... -.... ~··r ., ,

.n .. m .. • .. ••• .. ··. &;;i; .. ·~~~· ·· __ -\O ·S 10 IS Xl l5

los(LRmln) For ,'iIoP!\ln& LR - 0 "'ltr>td lotIUl)_ -S

TrueAllele (No locus LR threshold) I 0 8o!l!Amps DAmpI AAmpl

'¥ '" •

i 0

~ • 'ff 8

-; • E " ~ • J

" ~ • . " ;·1 ···· g }·~~io ~ ·· ··· ih D ··;;··· ., 5 10 IS

loJ1LRmin) For ."phI .... 1ft • 0 n fllntd l.OI1UlJ . -s

~t\. /II.-( Steven P. Myer~\ --z, I ,,,

" " '" lo,(LRmln) For .... phl"" Ul . 0 IIslant<l lOlIIRj - -S

TrueAllele (Locus LR threshold = 0.01)

" • " " ..

• , ~ •

a • • a~ ":";fJljAW~:.,a~i '!l 'li ih D . ., S 10 IS

lol1lRmlnl for .,,!)hI .... Ul • 0 IIIItMd lOIMJ • -5

"

Page 2 1 of25 epic.org EPIC-16-02-02-CalDOJ-FOIA-20160219-Validation-Study-Diff-Degradation 000021



December 10, 20 14

• QCO 18 (male, degradation series)

STRmix (V 1.0.7.49 allele variance) 0 80thAmps DAmp I .... '

• '"

~

~ " 5

! J i w E 5 J ,

~--Q--O

.............. Ji, ........ ~.: .................. .......... ~.

·w . , • , " 1000LRmlni

For.r~phI .... L1\.OUlilMdlOl(LII.I_ ·5

TrueAllele (No locus LR threshold)

o

• " •

" 8

" '" "

STRmix (V2.0 allele variance) OIllolllAmPt DAmp 1 to Amp2

'"

.-~ " J i w

~ J ,

. ........................ ...... iii: ... ~ __ .......... • ,

·w ., • , w " '" " lollLRmin} For ."plll ..... UI • 0 nsltntd LOIIIRI. ·5

TrueAllele (Locus LR threshold ~ 0.0 I) o 8o\hAmps C AmpI !l Ampl

", I

o o 8 0 ' 0

0 0 0

a~-_

'" "

............ ~ ... ~, .. ~ .~t+.~ ... ,\ .. O .... _ .. ~h .. . ~-BO-----' 5 10 IS ., loc(LRmln)

FOI ... pNflI,lft • 0 nl!ano<1lOlILR)- ·5

PRECISION: Graphing the 3-person mixture • See the fo llowing spreadsheets

o STRmix 3PM_DifDeg Graphs (spm 06.24.2014).xlsm

, " lollLRmln)

Forl'itPhlrc.1R .O .... M<ll .. IUlI··5

o TAC 3PM_DifDeg 0 and 0. 01 Graphs (spm 06.24.20 14).xlsm • LRH within like interpretations were compared in a pairwise manner.

15 20 15

o E.g. , the results of the th ree joint interpretations for a mixture were compared to each other. • The results were plotted as the lower of the two log,o LRH on the x-axis and the absolute value of the

difference on the y-axis. o The dashed line represents I log unit, i.e. , a factor of 10 difference. o LRH ~ 0 was plotted as - 5.

~ /\...f Steven P Myers /'j7' 1'1 II ~ l of25 epic.org EPIC-16-02-02-CalDOJ-FOIA-20160219-Validation-Study-Diff-Degradation 000022


• STRmix

o o

Casework Interpretation of Complex DNA Mixtures PART I: Differential degradat ion studies

o 80th Amps 0 Amp 1 II Amp 2

14

12

10

8

• 4

2

~ ___ ",_",_" ,_,u~u l "'" .u ... uu .. Q ... uu,~~=-~_t. u "u",uu

· 10 ·5 o 5 10 IS 20 25

Log(LRmin) For graphing, lR . 0 assigned Log(lR)·-S

• TrueAllele Locus LR threshold ~ 0.0 1 o Both Amps DAmpl Il Amp 2

14

12

10

8

• 4

<> <>

~···_··_···_··_···_···~·· __ ·_··_···_·:, ~· I_···_··JU;~.~:~.~~.~_. ____ ~~~_ .. _ ... _ .. _ ... ~ ... _~~~. __ ~ __ ~ ·10 ·s o 5 10 IS 20 2S

Log(LRmin) For graphing.lR . 0 assigned log(lR) - -5

Steven P Myer~\ ,0 (I '-'l

December 10, 2014




• TrueAllele (No locus LR threshold) o 80th Amps DAmpl 6 Amp2

14

12

10

8

6

4

t; ................... ......... ....... . Ill! ... -0; ./5, ...... 4\. ~--__ IO ____ 04-~n~~ ______ ~A ________ t; __ ~ ____ ~

· 10 ·s o 5 10 15 20 25

log(lRmin) For graphing, lR .. 0 assigned log(LR) .. -5

PRECISION: Summary Tables • See the following spreadsheets

o STRmix_Graphs_Data (spm 07.10.2014).xlsm o TAC_Graphs_Data (spm 07. 10.2014).xlsm

• Notes:

December 10,2014

~.;QW\ o These tables do not include the results from the 2-person mixture differential degradation j study's

"QCI8-QC I08_intact" mixture. o STRmix results us ing the V 1.0.7.49 variance were not included.

• All contributors for 3 replicates each of amp I, amp 2, and the joint interpretation:

Total Min Max %> I


2.9E-05 7.3E-03 2.9E-05 9.6E-0 I 2.3E+OI 2.3E+OI 0.00% 11.11% 1.59%

TAC (0.0 1 minimum) 2PM 3PM Comb 162 27 189

O.OE+OO 2.2E-02 O.OE+OO 1.3E+OI 3.4E+00 l.3E+OI 46.30% 40.74% 45.50%

• All contributors for 3 replicates each of amp I and amp 2:

Total Min Max %> I


2.9E-05 7.3E-03 2.9E-05 5.6E-O I 2.3E+O I 2.3E+O I 0.00% 16.67% 2.38%

Steven P Myers ~IO Ilt.{


O.OE+OO 5.4E-02 O.OE+OO 1.3E+OI 1.8E+00 1.3E+OI 40.74% 27.78% 38.89%


O.OE+OO O.OE+OO O.OE+OO 2 .7E+01 2.0E+00 2.7E+OI 40.12% 22.22% 37.57%


O.OE+OO O.OE+OO O.OE+OO 2.3E+OI 2.0E+00 2.3E+OI 43.52% 33.33% 42.06%




December 10,2014

• All contributors for 3 replicates each of the joint interpretations:

STRmix (V2.0 variance) TAC (0.01 minimum) TAC (0 minimum) 2PM 3PM Comb 2PM 3PM Comb 2PM 3PM Comb

54 9 63 54 9 63 54 9 63 Total Min Max % > I

I.O E-02 1.4E-02 I.OE-02 O.OE+OO 2.2E-02 O.OE+OO O.OE+OO O.OE+OO O.OE+OO 9.6E-OI 3.6E-OI 9.6E-OI 1.1 HO I 3.4E+00 I.IE+OI 2.7HOI O.OE+OO 0.00% 0.00% 0.00% 57.4 1% 66.67% 58.73% 33.33% 0.00%

Discussion and Conclusions

Sensitivity was measured as the proportion of comparisons (including each mixture being compared to each contributor) that gave positive 10g(LRH)'

2.7E+OI 28.57%

MixMaster 1. 1 sensitivity was 97.22% for the 2-person mixtures when using the standard 4-allele approach to estimating Mx. The remaining 2.78% was attributed to a single false exclusion (LR ~ 0) at a single locus for the comparison of the male reference (QC I08) to one amplification of the sample with the most degraded male DNA (Tube 16 Amp 1). The female profile (QCO 18) was inc luded in the results for this mixture. When MixMaster was rerun using the Mx Calculator 3 and 4-allele Mx estimate, this comparison was no longer LR ~ O.

STRmix sensitivity was 100% for the 2-person mixtures, and 96.3% for the 3-person mixture when using the V2.0 variance supplied by ESR. Overall, STRmix had positive 10g(LRH) in 99.47% of the comparisons. The remaining 0.53% was attributed to a single false exclusion (LR ~ 0) for the degraded DNA contributor to the 3-person mixture. In nine-total comparisons to this person, the false exclusion occurred in one interpretation of Amp 1. The other two interpretations of Amp I, all three interpretations of Amp 2, and all three joint interpretations gave positive 10g(LRH)' Had the V 1.0.7.49 Model Maker variance been the only option, sensitivity would have dropped considerably, as evidence by the false negatives (LR ~ 0) observed in the graphs.

TrueAllele Casework sensitivity was 82.1 % for the 2-person mixtures, and 100% for the 3-person mixture. Overall, TrueAllele had positive 10g(LRH) in 84.66% of the comparisons. However, when the minimum LR was bypassed (i.e., LR ~ 0 allowed), the sensitivity decreased to 39.51 % for the 2-person mixtures, and 44.44% for the 3-person mixture. Overall, when bypassing the minimum locus LR threshold, sensitivity was only 40.21 %, with over half (56.08%) of the comparisons giving LR ~ 0 for at least one locus.

Precision was measured as the proportion of pairwise comparisons, within like-interpretations, that were within one 10g(LRH) unit.

STRmix precision was 100% for the 2-person mixtures, and 88.89% for the 3-person mixture when using the V2.0 variance supplied by ESR. Overall, STRmix had 98.41 % of the pairwise comparisons within one log unit. The remaining 1.59% was attributed to the single false exclusion (LR ~ 0) for the degraded DNA contributor to the 3-person mixture. This one false exclusion was compared to two other interpretations that have positive 10g(LRH)'

TrueAllele Casework precision was 53.7% for the 2-person mixtures, and 59.26% for the 3-person mixture. Overall, TrueAllele had 54.5% of the pairwise comparisons within one log unit. Paradoxically, this appeared to improve when the minimum locus LR was bypassed. However, this appears to be an artifact of the increased number ofLR ~ O. When comparing two false negatives, the LRs are identical, i.e., the LR ~ O.

Overall , STRmix had better sensitivity and precision than TrueAllele Casework for the same number of interpretations. (STRmix also outperformed MixMaster (standard settings) for the interpretation of the 2-person set.) The degradation approach in STRmix appeared to better handle the kind of differential degradation highlighted in these samples - where the Mx of the contributors flips in the middle of the locus size (base pair) range.

Steven P Myers 40 !t~ of25 epic.org EPIC-16-02-02-CalDOJ-FOIA-20160219-Validation-Study-Diff-Degradation 000025

strmix vi.0.7.49 truealiele casework purpose...• strmix is a stand-alone program written in java...

Documents