![Page 1: Understanding PCR - Biostatistics · enzymes don’t know the origin of the template 13. ... Fusion 30 cycles SGMPlus 34 cycles. 50 Mock samples v Casework There is a strong propensity](https://reader033.vdocuments.us/reader033/viewer/2022041912/5e6810d4a7de196c58568bff/html5/thumbnails/1.jpg)
Understanding PCR
• Total allelic product• Peak height • Back stutter• Forward stutter• Variability in peak height
1
![Page 2: Understanding PCR - Biostatistics · enzymes don’t know the origin of the template 13. ... Fusion 30 cycles SGMPlus 34 cycles. 50 Mock samples v Casework There is a strong propensity](https://reader033.vdocuments.us/reader033/viewer/2022041912/5e6810d4a7de196c58568bff/html5/thumbnails/2.jpg)
Total allelic product (TAP)
• Total allelic product, Ta = height of allele plus height of stutter peaks
• Ta=Oa-1+Oa+Oa+1
2
Allelea
a-1 a+1Forward stutter
Back stutter
![Page 3: Understanding PCR - Biostatistics · enzymes don’t know the origin of the template 13. ... Fusion 30 cycles SGMPlus 34 cycles. 50 Mock samples v Casework There is a strong propensity](https://reader033.vdocuments.us/reader033/viewer/2022041912/5e6810d4a7de196c58568bff/html5/thumbnails/3.jpg)
Effects on TAP
• Template• Degradation• Locus specific effects• Additivity
3
![Page 4: Understanding PCR - Biostatistics · enzymes don’t know the origin of the template 13. ... Fusion 30 cycles SGMPlus 34 cycles. 50 Mock samples v Casework There is a strong propensity](https://reader033.vdocuments.us/reader033/viewer/2022041912/5e6810d4a7de196c58568bff/html5/thumbnails/4.jpg)
4
1
10
100
1000
10000
1 10 100 1000aver
age
peak
hei
ght (
rfu)
amount of DNA added to PCR (pg)
![Page 5: Understanding PCR - Biostatistics · enzymes don’t know the origin of the template 13. ... Fusion 30 cycles SGMPlus 34 cycles. 50 Mock samples v Casework There is a strong propensity](https://reader033.vdocuments.us/reader033/viewer/2022041912/5e6810d4a7de196c58568bff/html5/thumbnails/5.jpg)
5
Profile slopes/degradation
![Page 6: Understanding PCR - Biostatistics · enzymes don’t know the origin of the template 13. ... Fusion 30 cycles SGMPlus 34 cycles. 50 Mock samples v Casework There is a strong propensity](https://reader033.vdocuments.us/reader033/viewer/2022041912/5e6810d4a7de196c58568bff/html5/thumbnails/6.jpg)
Theory
6
• Consider that the degradation of the DNA strand was random with respect to location. • Consider a fragment of length l. • If the probability of a break is p, at any of the
locations 1…l• the chance of the full fragment being amplified is
(1-p)l. • Since 1-p is less than 1 this equation describes an exponential decline in peak height.
![Page 7: Understanding PCR - Biostatistics · enzymes don’t know the origin of the template 13. ... Fusion 30 cycles SGMPlus 34 cycles. 50 Mock samples v Casework There is a strong propensity](https://reader033.vdocuments.us/reader033/viewer/2022041912/5e6810d4a7de196c58568bff/html5/thumbnails/7.jpg)
7
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
70 120 170 220 270 320 370 420
prob
ability
Fragment length
Bacteria, UV, others
Positive
Negative
Positive
![Page 8: Understanding PCR - Biostatistics · enzymes don’t know the origin of the template 13. ... Fusion 30 cycles SGMPlus 34 cycles. 50 Mock samples v Casework There is a strong propensity](https://reader033.vdocuments.us/reader033/viewer/2022041912/5e6810d4a7de196c58568bff/html5/thumbnails/8.jpg)
8
‐1.5
‐1.0
‐0.5
0.0
0.5
1.0
1.5
2.0
70 170 270 370
log(Oa/E a)
ma
![Page 9: Understanding PCR - Biostatistics · enzymes don’t know the origin of the template 13. ... Fusion 30 cycles SGMPlus 34 cycles. 50 Mock samples v Casework There is a strong propensity](https://reader033.vdocuments.us/reader033/viewer/2022041912/5e6810d4a7de196c58568bff/html5/thumbnails/9.jpg)
9
‐1.5
‐1.0
‐0.5
0.0
0.5
1.0
1.5
2.0
70 170 270 370
log(Oa/E a)
ma
Linear Exponential
![Page 10: Understanding PCR - Biostatistics · enzymes don’t know the origin of the template 13. ... Fusion 30 cycles SGMPlus 34 cycles. 50 Mock samples v Casework There is a strong propensity](https://reader033.vdocuments.us/reader033/viewer/2022041912/5e6810d4a7de196c58568bff/html5/thumbnails/10.jpg)
10
4. Locus specific amplification
0
200
400
600
800
1000
1200
1400
1600
1800
2000
70 120 170 220 270 320 370 420
![Page 11: Understanding PCR - Biostatistics · enzymes don’t know the origin of the template 13. ... Fusion 30 cycles SGMPlus 34 cycles. 50 Mock samples v Casework There is a strong propensity](https://reader033.vdocuments.us/reader033/viewer/2022041912/5e6810d4a7de196c58568bff/html5/thumbnails/11.jpg)
4. Locus specific amplification
• Observation that some loci amplify more efficiently than others
• This effect varies with time• Results in varying peak heights off the
general trend• Locus offset at each locus allows for this
variation
11
![Page 12: Understanding PCR - Biostatistics · enzymes don’t know the origin of the template 13. ... Fusion 30 cycles SGMPlus 34 cycles. 50 Mock samples v Casework There is a strong propensity](https://reader033.vdocuments.us/reader033/viewer/2022041912/5e6810d4a7de196c58568bff/html5/thumbnails/12.jpg)
Locus specific amplification example
12
![Page 13: Understanding PCR - Biostatistics · enzymes don’t know the origin of the template 13. ... Fusion 30 cycles SGMPlus 34 cycles. 50 Mock samples v Casework There is a strong propensity](https://reader033.vdocuments.us/reader033/viewer/2022041912/5e6810d4a7de196c58568bff/html5/thumbnails/13.jpg)
Assumption of additivity
• For many years we have assumed that the contribution from various sources adds
• Often termed stacking in the US,• Allele + allele, allele + stutter• Logical, based on the fact that the
enzymes don’t know the origin of the template
13
![Page 14: Understanding PCR - Biostatistics · enzymes don’t know the origin of the template 13. ... Fusion 30 cycles SGMPlus 34 cycles. 50 Mock samples v Casework There is a strong propensity](https://reader033.vdocuments.us/reader033/viewer/2022041912/5e6810d4a7de196c58568bff/html5/thumbnails/14.jpg)
Recent questioning of additivity
14
Keith Inman, Norah Rudin, Kirk LohmuellerCalifornia State University East Bay
Presented at California Association of CriminalistsMeeting 2017
..if the relative contributions are not additive, then that calculation is not supportable.
To determine if this practice is scientifically supportable, it would be useful to query a large data set of mixtures created from known profiles, designed specifically to answer this question.
![Page 15: Understanding PCR - Biostatistics · enzymes don’t know the origin of the template 13. ... Fusion 30 cycles SGMPlus 34 cycles. 50 Mock samples v Casework There is a strong propensity](https://reader033.vdocuments.us/reader033/viewer/2022041912/5e6810d4a7de196c58568bff/html5/thumbnails/15.jpg)
15
‐0.5
‐0.4
‐0.3
‐0.2
‐0.1
0.0
0.1
0.2
0.3
0.4
0.5
0 1000 2000 3000 4000 5000 6000
log(O/E)
Expected height of peak
loglala
OE
vs laE
allele
Expected heights developed by fitting a model of template and degradation
Perfect would be all points on 0.0Variance higher at low peak height (see x axis)
![Page 16: Understanding PCR - Biostatistics · enzymes don’t know the origin of the template 13. ... Fusion 30 cycles SGMPlus 34 cycles. 50 Mock samples v Casework There is a strong propensity](https://reader033.vdocuments.us/reader033/viewer/2022041912/5e6810d4a7de196c58568bff/html5/thumbnails/16.jpg)
16
‐0.5
‐0.4
‐0.3
‐0.2
‐0.1
0.0
0.1
0.2
0.3
0.4
0.5
0 1000 2000 3000 4000 5000 6000
log(O/E)
Expected height of peak
‐0.5
‐0.4
‐0.3
‐0.2
‐0.1
0.0
0.1
0.2
0.3
0.4
0.5
0 1000 2000 3000 4000 5000 6000log(O/E)
Expected height of peak
allele is minor allele non‐minor‐0.5
‐0.4
‐0.3
‐0.2
‐0.1
0.0
0.1
0.2
0.3
0.4
0.5
0 1000 2000 3000 4000 5000 6000
log(O/E)
Expected height of peak
loglala
OE
vs laE
allele stuttercomposite
Expected heights developed by fitting a model of template and degradation
![Page 17: Understanding PCR - Biostatistics · enzymes don’t know the origin of the template 13. ... Fusion 30 cycles SGMPlus 34 cycles. 50 Mock samples v Casework There is a strong propensity](https://reader033.vdocuments.us/reader033/viewer/2022041912/5e6810d4a7de196c58568bff/html5/thumbnails/17.jpg)
Stutter• By-product of PCR process • Generally one repeat unit smaller than
target alleleAn epg of a heterozygote D19S433 13,15.2 displaying stutter peaks one repeat unit less than the parent alleles
![Page 18: Understanding PCR - Biostatistics · enzymes don’t know the origin of the template 13. ... Fusion 30 cycles SGMPlus 34 cycles. 50 Mock samples v Casework There is a strong propensity](https://reader033.vdocuments.us/reader033/viewer/2022041912/5e6810d4a7de196c58568bff/html5/thumbnails/18.jpg)
Stutter mechanism• A proposed mechanism is slipped strand
mispairing (SSM).• During PCR
– the DNA polymerase enzyme stalls, – dissociates from the DNA, – the template strand “loops out”, – and the new strand is one repeat unit shorter
than the template strand
![Page 19: Understanding PCR - Biostatistics · enzymes don’t know the origin of the template 13. ... Fusion 30 cycles SGMPlus 34 cycles. 50 Mock samples v Casework There is a strong propensity](https://reader033.vdocuments.us/reader033/viewer/2022041912/5e6810d4a7de196c58568bff/html5/thumbnails/19.jpg)
Stutter rates• Degree of stutter formation is related to
the type of repeat• STRs with di- and trinucleotide repeat
structures are known to stutter more than tetra- and pentanucleotide repeats
• Most forensic markers are tetranucleotide repeats
AATG AATG AATG AATG AATG AATGPrimer Primer
Flanking region
Flanking region
STR
![Page 20: Understanding PCR - Biostatistics · enzymes don’t know the origin of the template 13. ... Fusion 30 cycles SGMPlus 34 cycles. 50 Mock samples v Casework There is a strong propensity](https://reader033.vdocuments.us/reader033/viewer/2022041912/5e6810d4a7de196c58568bff/html5/thumbnails/20.jpg)
Summary (& additional forms of stutter)Type of stutter Possible mechanism
Back stutter (minus one full repeat)
Slipped strand mis‐pairing (SSM)
Forward stutter (plus one full repeat)
A loop of one repeat units forms in the nascent strand, resulting in insertion of one repeat
Double back stutter (minus 2 full repeats)
A loop of two repeat units forms in template strand resulting in deletion of two repeats (some new evidence suggesting this might be two single stutters)
Minus 2 base pair SSM but of 2 b repeat within the allele (observed as well as 4 base pair repeat)
![Page 21: Understanding PCR - Biostatistics · enzymes don’t know the origin of the template 13. ... Fusion 30 cycles SGMPlus 34 cycles. 50 Mock samples v Casework There is a strong propensity](https://reader033.vdocuments.us/reader033/viewer/2022041912/5e6810d4a7de196c58568bff/html5/thumbnails/21.jpg)
21
Nascent strand
Template strand
Or breathing
![Page 22: Understanding PCR - Biostatistics · enzymes don’t know the origin of the template 13. ... Fusion 30 cycles SGMPlus 34 cycles. 50 Mock samples v Casework There is a strong propensity](https://reader033.vdocuments.us/reader033/viewer/2022041912/5e6810d4a7de196c58568bff/html5/thumbnails/22.jpg)
Stutter ratios• Back stutter is typically quantified as a
stutter ratio (SR):
• where Oa-1 refers to the observed height of the stutter peak, and Oa the parent peak
1a
a
OSRO
![Page 23: Understanding PCR - Biostatistics · enzymes don’t know the origin of the template 13. ... Fusion 30 cycles SGMPlus 34 cycles. 50 Mock samples v Casework There is a strong propensity](https://reader033.vdocuments.us/reader033/viewer/2022041912/5e6810d4a7de196c58568bff/html5/thumbnails/23.jpg)
Stutter ratios• Forward stutter is typically quantified as a
stutter ratio (FSR):
• where Oa+1 refers to the observed height of the stutter peak, and Oa the parent peak
1a
aa
OFSRO
![Page 24: Understanding PCR - Biostatistics · enzymes don’t know the origin of the template 13. ... Fusion 30 cycles SGMPlus 34 cycles. 50 Mock samples v Casework There is a strong propensity](https://reader033.vdocuments.us/reader033/viewer/2022041912/5e6810d4a7de196c58568bff/html5/thumbnails/24.jpg)
Stutter ratios• Traditionally we apply a filter at analysis to
remove stutter e.g. 15%– Locus specific– Profile/Kit specific
• Stutter filters remove the label, but not the peak
![Page 25: Understanding PCR - Biostatistics · enzymes don’t know the origin of the template 13. ... Fusion 30 cycles SGMPlus 34 cycles. 50 Mock samples v Casework There is a strong propensity](https://reader033.vdocuments.us/reader033/viewer/2022041912/5e6810d4a7de196c58568bff/html5/thumbnails/25.jpg)
25
![Page 26: Understanding PCR - Biostatistics · enzymes don’t know the origin of the template 13. ... Fusion 30 cycles SGMPlus 34 cycles. 50 Mock samples v Casework There is a strong propensity](https://reader033.vdocuments.us/reader033/viewer/2022041912/5e6810d4a7de196c58568bff/html5/thumbnails/26.jpg)
Stutter ratios• This is an in/out (binary) decision• Most important when the minor POI is
approximately same height as stutter• Such a peak may be stutter or stutter/allelic• Making an in/out decision may be incorrect and
have consequences
![Page 27: Understanding PCR - Biostatistics · enzymes don’t know the origin of the template 13. ... Fusion 30 cycles SGMPlus 34 cycles. 50 Mock samples v Casework There is a strong propensity](https://reader033.vdocuments.us/reader033/viewer/2022041912/5e6810d4a7de196c58568bff/html5/thumbnails/27.jpg)
Stutter products• The products are allelic in all aspects • Not distinguishable from true allelic
products• Contribute to the complexity of profile
interpretation• Especially when a true contributor’s alleles
are approximately the same height as the stutter product from another contributor.
![Page 28: Understanding PCR - Biostatistics · enzymes don’t know the origin of the template 13. ... Fusion 30 cycles SGMPlus 34 cycles. 50 Mock samples v Casework There is a strong propensity](https://reader033.vdocuments.us/reader033/viewer/2022041912/5e6810d4a7de196c58568bff/html5/thumbnails/28.jpg)
Stutter ratios • Stutter ratios are actually allele specific
![Page 29: Understanding PCR - Biostatistics · enzymes don’t know the origin of the template 13. ... Fusion 30 cycles SGMPlus 34 cycles. 50 Mock samples v Casework There is a strong propensity](https://reader033.vdocuments.us/reader033/viewer/2022041912/5e6810d4a7de196c58568bff/html5/thumbnails/29.jpg)
TH01 stutter
29
![Page 30: Understanding PCR - Biostatistics · enzymes don’t know the origin of the template 13. ... Fusion 30 cycles SGMPlus 34 cycles. 50 Mock samples v Casework There is a strong propensity](https://reader033.vdocuments.us/reader033/viewer/2022041912/5e6810d4a7de196c58568bff/html5/thumbnails/30.jpg)
TH01 repeat structure
Longest uninterrupted stretch (LUS) of basic repeat motifs is a good predictor of stutter ratio
Common TH01 allele sequencesRepeat structure Allele
[AATG]6 6[AATG]7 7[AATG]8 8[AATG]9 9
[AATG]6ATG[AATG]3 9.3
LUS67896
30
![Page 31: Understanding PCR - Biostatistics · enzymes don’t know the origin of the template 13. ... Fusion 30 cycles SGMPlus 34 cycles. 50 Mock samples v Casework There is a strong propensity](https://reader033.vdocuments.us/reader033/viewer/2022041912/5e6810d4a7de196c58568bff/html5/thumbnails/31.jpg)
TH01 Stutter ratio versus LUS
31
![Page 32: Understanding PCR - Biostatistics · enzymes don’t know the origin of the template 13. ... Fusion 30 cycles SGMPlus 34 cycles. 50 Mock samples v Casework There is a strong propensity](https://reader033.vdocuments.us/reader033/viewer/2022041912/5e6810d4a7de196c58568bff/html5/thumbnails/32.jpg)
Allele versus LUS, NGM Select loci
0.00
0.02
0.04
0.06
0.08
0.10
0.12
5 10 15 20 25 30 35
Stutter ratio
Allele
R2 = 27%
y = 0.0061x ‐ 0.0189
5 10 15 20 25
LUS
R2 = 61%
32
R-squared is a statistical measure of how close the data are to the fitted regression line
![Page 33: Understanding PCR - Biostatistics · enzymes don’t know the origin of the template 13. ... Fusion 30 cycles SGMPlus 34 cycles. 50 Mock samples v Casework There is a strong propensity](https://reader033.vdocuments.us/reader033/viewer/2022041912/5e6810d4a7de196c58568bff/html5/thumbnails/33.jpg)
33
21.2 [AAAG]2 AG [AAAG]3 AG [AAAG]9 AA AAAG [AAAG]11 G AAGG [AAAG]2 AG
21.2 [AAAG]2 AG [AAAG]3 AG [AAAG]11 AA AAAG [AAAG]9 G AAGG [AAAG]2 AG
22 [AAAG]2 AG [AAAG]3 AG [AAAG]22 G [AAAG]3 AG
22.2 [AAAG]2 AG [AAAG]3 AG [AAAG]7 AA AAAG [AAAG]14 G AAGG [AAAG]2 AG
22.2 [AAAG]2 AG [AAAG]3 AG [AAAG]8 [AG]5 [AAAG]12 G AAGG [AAAG]2 AG
22.2 [AAAG]2 AG [AAAG]3 AG [AAAG]9 AA AAAG [AAAG]12 G AAGG [AAAG]2 AG
![Page 34: Understanding PCR - Biostatistics · enzymes don’t know the origin of the template 13. ... Fusion 30 cycles SGMPlus 34 cycles. 50 Mock samples v Casework There is a strong propensity](https://reader033.vdocuments.us/reader033/viewer/2022041912/5e6810d4a7de196c58568bff/html5/thumbnails/34.jpg)
34
‐0.05
‐0.03
‐0.01
0.01
0.03
0.05
10 15 20 25 30 35
O-E
Allele
max 5.83,0ii
SR m l c
![Page 35: Understanding PCR - Biostatistics · enzymes don’t know the origin of the template 13. ... Fusion 30 cycles SGMPlus 34 cycles. 50 Mock samples v Casework There is a strong propensity](https://reader033.vdocuments.us/reader033/viewer/2022041912/5e6810d4a7de196c58568bff/html5/thumbnails/35.jpg)
35
21.2 [AAAG]2 AG [AAAG]3 AG [AAAG]9 AA AAAG [AAAG]11 G AAGG [AAAG]2 AG
21.2 [AAAG]2 AG [AAAG]3 AG [AAAG]11 AA AAAG [AAAG]9 G AAGG [AAAG]2 AG
22 [AAAG]2 AG [AAAG]3 AG [AAAG]22 G [AAAG]3 AG
22.2 [AAAG]2 AG [AAAG]3 AG [AAAG]7 AA AAAG [AAAG]14 G AAGG [AAAG]2 AG
22.2 [AAAG]2 AG [AAAG]3 AG [AAAG]8 [AG]5 [AAAG]12 G AAGG [AAAG]2 AG
22.2 [AAAG]2 AG [AAAG]3 AG [AAAG]9 AA AAAG [AAAG]12 G AAGG [AAAG]2 AG
2 - 5.83 = 0
3 - 5.83 = 0
9 - 5.83 = 3.17
12 - 5.83 = 6.17
2 - 5.83 = 0
![Page 36: Understanding PCR - Biostatistics · enzymes don’t know the origin of the template 13. ... Fusion 30 cycles SGMPlus 34 cycles. 50 Mock samples v Casework There is a strong propensity](https://reader033.vdocuments.us/reader033/viewer/2022041912/5e6810d4a7de196c58568bff/html5/thumbnails/36.jpg)
36
![Page 37: Understanding PCR - Biostatistics · enzymes don’t know the origin of the template 13. ... Fusion 30 cycles SGMPlus 34 cycles. 50 Mock samples v Casework There is a strong propensity](https://reader033.vdocuments.us/reader033/viewer/2022041912/5e6810d4a7de196c58568bff/html5/thumbnails/37.jpg)
Stutter filters
3
Not FilteredSignificantlyOver Filtered
![Page 38: Understanding PCR - Biostatistics · enzymes don’t know the origin of the template 13. ... Fusion 30 cycles SGMPlus 34 cycles. 50 Mock samples v Casework There is a strong propensity](https://reader033.vdocuments.us/reader033/viewer/2022041912/5e6810d4a7de196c58568bff/html5/thumbnails/38.jpg)
3
y=0.008x – 0.092 + 2(0.0264)
![Page 39: Understanding PCR - Biostatistics · enzymes don’t know the origin of the template 13. ... Fusion 30 cycles SGMPlus 34 cycles. 50 Mock samples v Casework There is a strong propensity](https://reader033.vdocuments.us/reader033/viewer/2022041912/5e6810d4a7de196c58568bff/html5/thumbnails/39.jpg)
Forward stutter filters
39
![Page 40: Understanding PCR - Biostatistics · enzymes don’t know the origin of the template 13. ... Fusion 30 cycles SGMPlus 34 cycles. 50 Mock samples v Casework There is a strong propensity](https://reader033.vdocuments.us/reader033/viewer/2022041912/5e6810d4a7de196c58568bff/html5/thumbnails/40.jpg)
40
Locus FSR Locus FSRCSF1PO 0.0062 D2S441 0.0055D10S1248 0.0128 D3S1358 0.0042D12S391 0.0029 D5S818 0.0077D13S317 0.0046 D7S820 0.0020D16S539 0.0059 D8S1179 0.0054D18S51 0.0045 FGA 0.0030D19S433 0.0019 SE33 0.0059D1S1656 0.0052 TH01 0.0006D21S11 0.0072 TPOX 0.0007D2S1338 0.0016 vWA 0.0033
Forward stutter filtersusually per locus
![Page 41: Understanding PCR - Biostatistics · enzymes don’t know the origin of the template 13. ... Fusion 30 cycles SGMPlus 34 cycles. 50 Mock samples v Casework There is a strong propensity](https://reader033.vdocuments.us/reader033/viewer/2022041912/5e6810d4a7de196c58568bff/html5/thumbnails/41.jpg)
41
Forward stutter filtersexception is D22
Back stutter Forward stutter
![Page 42: Understanding PCR - Biostatistics · enzymes don’t know the origin of the template 13. ... Fusion 30 cycles SGMPlus 34 cycles. 50 Mock samples v Casework There is a strong propensity](https://reader033.vdocuments.us/reader033/viewer/2022041912/5e6810d4a7de196c58568bff/html5/thumbnails/42.jpg)
42
Double back stutter
![Page 43: Understanding PCR - Biostatistics · enzymes don’t know the origin of the template 13. ... Fusion 30 cycles SGMPlus 34 cycles. 50 Mock samples v Casework There is a strong propensity](https://reader033.vdocuments.us/reader033/viewer/2022041912/5e6810d4a7de196c58568bff/html5/thumbnails/43.jpg)
• This means you can no longer consider stutter filters on a simple “%” basis
• Find back stutter contribution• Find forward stutter contribution• Add them together
4
Combined Stutter
![Page 44: Understanding PCR - Biostatistics · enzymes don’t know the origin of the template 13. ... Fusion 30 cycles SGMPlus 34 cycles. 50 Mock samples v Casework There is a strong propensity](https://reader033.vdocuments.us/reader033/viewer/2022041912/5e6810d4a7de196c58568bff/html5/thumbnails/44.jpg)
4
Total Study Results64 2 person mixtures
Model Over Under After 3SD TotalNew 10 5 0 10Traditional 61 25 NA 86
54 3 person mixturesModel Over Under After 3SD TotalNew 7 7 4 11Traditional 19 31 NA 50
54 4 person mixturesModel Over Under After 3SD Total Over/UnderNew 5 5 4 9 0Traditional 19 27 NA 46 2
Grand total“New School” 30“Old School” 186
![Page 45: Understanding PCR - Biostatistics · enzymes don’t know the origin of the template 13. ... Fusion 30 cycles SGMPlus 34 cycles. 50 Mock samples v Casework There is a strong propensity](https://reader033.vdocuments.us/reader033/viewer/2022041912/5e6810d4a7de196c58568bff/html5/thumbnails/45.jpg)
1. Heterozygote balance
![Page 46: Understanding PCR - Biostatistics · enzymes don’t know the origin of the template 13. ... Fusion 30 cycles SGMPlus 34 cycles. 50 Mock samples v Casework There is a strong propensity](https://reader033.vdocuments.us/reader033/viewer/2022041912/5e6810d4a7de196c58568bff/html5/thumbnails/46.jpg)
Heterozygote balance• Also called peak height ratio• Ratio of two heterozygote peaks at a locus
• Two common definitions
1HMW
LMW
OHbO
smaller2
larger
OHb PHRO
Where O is observed peak height
![Page 47: Understanding PCR - Biostatistics · enzymes don’t know the origin of the template 13. ... Fusion 30 cycles SGMPlus 34 cycles. 50 Mock samples v Casework There is a strong propensity](https://reader033.vdocuments.us/reader033/viewer/2022041912/5e6810d4a7de196c58568bff/html5/thumbnails/47.jpg)
Heterozygote balance
1HMW
LMW
OHbO
1700 rfu
1790 rfu
smaller2
larger
OHb PHRO
17001790 0.95
• Hb1 has the highest information content because it maintains peak order
• Hb2 may be obtained from Hb1 but not vice versa
![Page 48: Understanding PCR - Biostatistics · enzymes don’t know the origin of the template 13. ... Fusion 30 cycles SGMPlus 34 cycles. 50 Mock samples v Casework There is a strong propensity](https://reader033.vdocuments.us/reader033/viewer/2022041912/5e6810d4a7de196c58568bff/html5/thumbnails/48.jpg)
Hb versus average peak height
“Most ratios look better as logs”
![Page 49: Understanding PCR - Biostatistics · enzymes don’t know the origin of the template 13. ... Fusion 30 cycles SGMPlus 34 cycles. 50 Mock samples v Casework There is a strong propensity](https://reader033.vdocuments.us/reader033/viewer/2022041912/5e6810d4a7de196c58568bff/html5/thumbnails/49.jpg)
49
Consistent across platforms
‐0.8
‐0.6
‐0.4
‐0.2
0
0.2
0.4
0.6
0 1000 2000 3000 4000 5000 6000 7000 8000
Log(Hb)
APH
‐0.5
‐0.4
‐0.3
‐0.2
‐0.1
0
0.1
0.2
0.3
0.4
0.5
0 1000 2000 3000 4000 5000 6000 7000 8000
Log(Hb
)
APH
‐0.6
‐0.4
‐0.2
0
0.2
0.4
0.6
0 500 1000 1500 2000 2500 3000 3500 4000
Log(Hb
)
APH
Identifiler 28 cycles GlobalFiler 29 cycles
Fusion 30 cyclesSGMPlus 34 cycles
![Page 50: Understanding PCR - Biostatistics · enzymes don’t know the origin of the template 13. ... Fusion 30 cycles SGMPlus 34 cycles. 50 Mock samples v Casework There is a strong propensity](https://reader033.vdocuments.us/reader033/viewer/2022041912/5e6810d4a7de196c58568bff/html5/thumbnails/50.jpg)
50
Mock samples v Casework
There is a strong propensity towards mock samples in the USBut some criticism that these do not model casework
![Page 51: Understanding PCR - Biostatistics · enzymes don’t know the origin of the template 13. ... Fusion 30 cycles SGMPlus 34 cycles. 50 Mock samples v Casework There is a strong propensity](https://reader033.vdocuments.us/reader033/viewer/2022041912/5e6810d4a7de196c58568bff/html5/thumbnails/51.jpg)
51
Mock single source Casework single sourceMock mixtures
![Page 52: Understanding PCR - Biostatistics · enzymes don’t know the origin of the template 13. ... Fusion 30 cycles SGMPlus 34 cycles. 50 Mock samples v Casework There is a strong propensity](https://reader033.vdocuments.us/reader033/viewer/2022041912/5e6810d4a7de196c58568bff/html5/thumbnails/52.jpg)
52
TransformThe variance of a sum is the sum of the variances (if independent)
log log
log log
HMW
LMW
HMW LMW
HHbHH H
var log var log var logHMW LMWHb H H
![Page 53: Understanding PCR - Biostatistics · enzymes don’t know the origin of the template 13. ... Fusion 30 cycles SGMPlus 34 cycles. 50 Mock samples v Casework There is a strong propensity](https://reader033.vdocuments.us/reader033/viewer/2022041912/5e6810d4a7de196c58568bff/html5/thumbnails/53.jpg)
53
Transformvar log var log var logHMW LMWHb H H
Assume var log var log var logHMW LMW iH H H
var log 2var log iHb H
![Page 54: Understanding PCR - Biostatistics · enzymes don’t know the origin of the template 13. ... Fusion 30 cycles SGMPlus 34 cycles. 50 Mock samples v Casework There is a strong propensity](https://reader033.vdocuments.us/reader033/viewer/2022041912/5e6810d4a7de196c58568bff/html5/thumbnails/54.jpg)
• the relative error on the peak areas is proportional to
54
N0 is the number of starting template0
1N
![Page 55: Understanding PCR - Biostatistics · enzymes don’t know the origin of the template 13. ... Fusion 30 cycles SGMPlus 34 cycles. 50 Mock samples v Casework There is a strong propensity](https://reader033.vdocuments.us/reader033/viewer/2022041912/5e6810d4a7de196c58568bff/html5/thumbnails/55.jpg)
What would a perfect model look like?
55
Log(
O/E
)
If we compare the observed profile to the expected profile…
Expected allele height
![Page 56: Understanding PCR - Biostatistics · enzymes don’t know the origin of the template 13. ... Fusion 30 cycles SGMPlus 34 cycles. 50 Mock samples v Casework There is a strong propensity](https://reader033.vdocuments.us/reader033/viewer/2022041912/5e6810d4a7de196c58568bff/html5/thumbnails/56.jpg)
What would a perfect model look like?
56
1E
![Page 57: Understanding PCR - Biostatistics · enzymes don’t know the origin of the template 13. ... Fusion 30 cycles SGMPlus 34 cycles. 50 Mock samples v Casework There is a strong propensity](https://reader033.vdocuments.us/reader033/viewer/2022041912/5e6810d4a7de196c58568bff/html5/thumbnails/57.jpg)
Variance of allele model• Mean ~ 0• Variance inversely
proportional to E
57
![Page 58: Understanding PCR - Biostatistics · enzymes don’t know the origin of the template 13. ... Fusion 30 cycles SGMPlus 34 cycles. 50 Mock samples v Casework There is a strong propensity](https://reader033.vdocuments.us/reader033/viewer/2022041912/5e6810d4a7de196c58568bff/html5/thumbnails/58.jpg)
Variance of stutter model• Mean ~ 0• Variance inversely
proportional to E
58
Void is because of drop-out of the stutter
![Page 59: Understanding PCR - Biostatistics · enzymes don’t know the origin of the template 13. ... Fusion 30 cycles SGMPlus 34 cycles. 50 Mock samples v Casework There is a strong propensity](https://reader033.vdocuments.us/reader033/viewer/2022041912/5e6810d4a7de196c58568bff/html5/thumbnails/59.jpg)
TM