Simulating and Modeling Extended Twin Family Data
description
Transcript of Simulating and Modeling Extended Twin Family Data
![Page 1: Simulating and Modeling Extended Twin Family Data](https://reader031.fdocuments.in/reader031/viewer/2022012916/568144fe550346895db1c95a/html5/thumbnails/1.jpg)
Simulating and Modeling Extended Twin Family Data
Matthew C. Keller
Sarah E. Medland
Hermine Maes
(also, Lindon Eaves, Mike Neale, Pete Hatemi, Laramie Duncan)
![Page 2: Simulating and Modeling Extended Twin Family Data](https://reader031.fdocuments.in/reader031/viewer/2022012916/568144fe550346895db1c95a/html5/thumbnails/2.jpg)
I.I. Motivation for using extended pedigreesMotivation for using extended pedigrees
- Briefly introduce the NTFD, Stealth, & - Briefly introduce the NTFD, Stealth, & Cascade modelsCascade models
II.II. Simulation using GeneEvolveSimulation using GeneEvolve
- Practical looking at changes in genetic - Practical looking at changes in genetic variance across timevariance across time
III.III. Use GeneEvolve to simulate extended twin family Use GeneEvolve to simulate extended twin family data & run in Mxdata & run in Mx
- Practical getting sensitivity analysis of CTD & - Practical getting sensitivity analysis of CTD & NTFDNTFD
Outline
![Page 3: Simulating and Modeling Extended Twin Family Data](https://reader031.fdocuments.in/reader031/viewer/2022012916/568144fe550346895db1c95a/html5/thumbnails/3.jpg)
Understand in general terms the reason for Understand in general terms the reason for extended twin family designsextended twin family designs
Learn how to use GeneEvolveLearn how to use GeneEvolve Understand how to derive biases & sampling Understand how to derive biases & sampling
distributions from simulationdistributions from simulation
Goals
![Page 4: Simulating and Modeling Extended Twin Family Data](https://reader031.fdocuments.in/reader031/viewer/2022012916/568144fe550346895db1c95a/html5/thumbnails/4.jpg)
Fully understand the logic, path diagrams, and Fully understand the logic, path diagrams, and scripting of extended twin family modelsscripting of extended twin family models
NON-Goal
![Page 5: Simulating and Modeling Extended Twin Family Data](https://reader031.fdocuments.in/reader031/viewer/2022012916/568144fe550346895db1c95a/html5/thumbnails/5.jpg)
SEM is great because…SEM is great because… Directs focus to effect sizes, not “significance” Directs focus to effect sizes, not “significance” Forces consideration of causes and consequencesForces consideration of causes and consequences Explicit disclosure of assumptionsExplicit disclosure of assumptions
Potential weakness…Potential weakness… Parameter reification: “Using the CTD we found that 50% Parameter reification: “Using the CTD we found that 50%
of variation is due to A and 20% to C.” of variation is due to A and 20% to C.”
Structural Equation Modeling (SEM) in BG
![Page 6: Simulating and Modeling Extended Twin Family Data](https://reader031.fdocuments.in/reader031/viewer/2022012916/568144fe550346895db1c95a/html5/thumbnails/6.jpg)
SEM is great because…SEM is great because… Directs focus to effect sizes, not “significance” Directs focus to effect sizes, not “significance” Forces consideration of causes and consequencesForces consideration of causes and consequences Explicit disclosure of assumptionsExplicit disclosure of assumptions
Potential weakness…Potential weakness… Parameter reification: “Using the CTD we found that 50% Parameter reification: “Using the CTD we found that 50%
of variation is due to A and 20% to C.” of variation is due to A and 20% to C.”
Structural Equation Modeling (SEM) in BG
Not necessarily. Only true under assumptions that may often be
unmet (e.g., D=0) and usually go untested. To the degree
assumptions wrong, estimates are biased.
![Page 7: Simulating and Modeling Extended Twin Family Data](https://reader031.fdocuments.in/reader031/viewer/2022012916/568144fe550346895db1c95a/html5/thumbnails/7.jpg)
PT1
Ca
Dd
Ee c
A
PT2
Ca
D d
Eec
A1/.25
1
Classical Twin Design (CTD)
![Page 8: Simulating and Modeling Extended Twin Family Data](https://reader031.fdocuments.in/reader031/viewer/2022012916/568144fe550346895db1c95a/html5/thumbnails/8.jpg)
PT1
Ca
Dd
Ee c
A
PT2
Ca
D d
Eec
A1/.25
1
Classical Twin Design (CTD) Assumption biased up biased downAssumption biased up biased down
Either D or C is zero A C & DEither D or C is zero A C & D
No assortative mating C DNo assortative mating C D
No A-C covariance C D & ANo A-C covariance C D & A
![Page 9: Simulating and Modeling Extended Twin Family Data](https://reader031.fdocuments.in/reader031/viewer/2022012916/568144fe550346895db1c95a/html5/thumbnails/9.jpg)
Why can’t we estimate A, C & D at same time using twins only?
• Solve the following two equations for A, C, & D:
CVmz = A + D + C
CVdz = 1/2A + 1/4D + C
![Page 10: Simulating and Modeling Extended Twin Family Data](https://reader031.fdocuments.in/reader031/viewer/2022012916/568144fe550346895db1c95a/html5/thumbnails/10.jpg)
Why does simply setting D or C to zero bias C & D down and A up?
• Information to estimate A comes from the ratio CVmz : 2*CVdz. The closer this ratio is to unity, the higher A is.
• If D & C both exist at the same time, D drives the ratio up, C drives it down. To the degree these effects ‘cancel each other out,’ it looks like A at the expense of D & C.
![Page 11: Simulating and Modeling Extended Twin Family Data](https://reader031.fdocuments.in/reader031/viewer/2022012916/568144fe550346895db1c95a/html5/thumbnails/11.jpg)
Adding parents gets us around these assumptions
Assumption biased up biased downAssumption biased up biased downEither D or C is zeroEither D or C is zero
No assortative matingNo assortative mating
No A-C covarianceNo A-C covariance
PMa
Ca
Dd
Ee c
A
qw
PFa
Ca
D d
Eec
A
qw
m
m
PT1
Ca
Dd
Ee c
A
PT2
Ca
Dd
Eec
A
mm
1/.25
µ
We don’t have to
make these
x x
![Page 12: Simulating and Modeling Extended Twin Family Data](https://reader031.fdocuments.in/reader031/viewer/2022012916/568144fe550346895db1c95a/html5/thumbnails/12.jpg)
With parents, we can break “C” up into:
S = env. factors shared only between sibs
F = familial env factors passed from parents to offspring
F
SC
PT1
SaDd
Ee s
A
f
F
PT2
S aD d
Ees
A
f
F1/.25
1
Parents also allow differentiation of C into S & F
PT1
Ca
Dd
Ee c
A
PT2
Ca
D d
Eec
A1/.25
1
![Page 13: Simulating and Modeling Extended Twin Family Data](https://reader031.fdocuments.in/reader031/viewer/2022012916/568144fe550346895db1c95a/html5/thumbnails/13.jpg)
Nuclear Twin Family Design (NTFD)
Assumptions:Assumptions: Only can estimate 3 of 4: A, D, S, and F (bias is variable)Only can estimate 3 of 4: A, D, S, and F (bias is variable) Assortative mating due to primary phenotypic assortment (bias is variable)Assortative mating due to primary phenotypic assortment (bias is variable)
Note: m estimated and
f fixed to 1
PMa
Sa
Dd
Ee s
A
q
x
w
f
F
PFa
Sa
D d
Ees
A
q
x
w
f
F
mm
PT1
Sa
Dd
Ee s
A
f
F
PT2
Sa
D d
Ees
A
f
F
mm
1/.25
1
µ
![Page 14: Simulating and Modeling Extended Twin Family Data](https://reader031.fdocuments.in/reader031/viewer/2022012916/568144fe550346895db1c95a/html5/thumbnails/14.jpg)
Stealth
Include twins and their sibs, parents, spouses, and Include twins and their sibs, parents, spouses, and offspring…offspring… Gives 17 unique covariances (MZ, DZ, Sib, P-O, Spousal, Gives 17 unique covariances (MZ, DZ, Sib, P-O, Spousal,
MZ avunc, DZ avunc, MZ cous, DZ cous, GP-GO, and 7 in-MZ avunc, DZ avunc, MZ cous, DZ cous, GP-GO, and 7 in-laws) laws)
88 covariances with sex effects88 covariances with sex effects
![Page 15: Simulating and Modeling Extended Twin Family Data](https://reader031.fdocuments.in/reader031/viewer/2022012916/568144fe550346895db1c95a/html5/thumbnails/15.jpg)
can be estimated simultaneously
PT1
Sa
Dd
Ee s
A
f
F
PT2
Sa
D d
Ees
A
f
F1/.25
1
Additional obs. covs with Stealth allow estimation of A, S, D, & F
d
(Remember: we’re not just estimating more effects. More importantly, we’re
reducing the bias in estimated effects!)
FS DA
![Page 16: Simulating and Modeling Extended Twin Family Data](https://reader031.fdocuments.in/reader031/viewer/2022012916/568144fe550346895db1c95a/html5/thumbnails/16.jpg)
Stealth
PMa
Sa
Dd
Ee s
A
q
x
w
f
F
PFa
Sa
D d
Ees
A
q
x
w
f
F
mm
PT1
Sa
Dd
Ee s
A
f
F
PMa
Sa
Dd
Ee s
A
q
x
w
f
F
PT2
Sa
D d
Ees
A
f
F
PFa
Sa
D d
Ees
A
q
x
w
f
F
mm
PCh
Sa
D d
Ees
A
f
F
mm
mm
PCh
Sa
Dd
Ee s
A
f
F
1/.25
1
µ
µ
µ
![Page 17: Simulating and Modeling Extended Twin Family Data](https://reader031.fdocuments.in/reader031/viewer/2022012916/568144fe550346895db1c95a/html5/thumbnails/17.jpg)
Stealth
Assumption biased up biased downAssumption biased up biased downPrimary assortative mating A, D, or F A, D, or FPrimary assortative mating A, D, or F A, D, or F
Primary AM: mates choose each other based on Primary AM: mates choose each other based on phenotypic similarityphenotypic similarity
Social homogamy: mates choose each other due to Social homogamy: mates choose each other due to environmental similarity (e.g., religion)environmental similarity (e.g., religion)
Convergence: mates become more similar over Convergence: mates become more similar over timetime
![Page 18: Simulating and Modeling Extended Twin Family Data](https://reader031.fdocuments.in/reader031/viewer/2022012916/568144fe550346895db1c95a/html5/thumbnails/18.jpg)
PMa
Sa
Dd
Ee
PMa
s
e
a
A
q
x
wf
f
~
~
~~d~s~
F
µ
PFa
Sa
D d
Ee
PFa
s
e
a
A
q
x
wf
f
~
~
~~
d~ s~
F
mm
PT1
Sa
Dd
Ee
PT1
s
e
a
Af
f
~
~
~~
d~s~F
PMa
Sa
Dd
Ee
PSp
s
e
a
A
q
x
wf
f
~
~
~~d~s~
F
PT2
Sa
D d
Ee
PT2
s
e
a
Af
f
~
~
~~
d~s~ F
PFa
Sa
D d
Ee
PSp
s
e
a
A
q
x
wf
f
~
~
~~
d~ s~
F
mm
PCh
Sa
D d
Ees
A
f
F
µ µ
mm
mm
PCh
Sa
Dd
Ee s
A
f
F
1/.25
1
Cascade
![Page 19: Simulating and Modeling Extended Twin Family Data](https://reader031.fdocuments.in/reader031/viewer/2022012916/568144fe550346895db1c95a/html5/thumbnails/19.jpg)
P
Ma
SaDd
Ee
P
Ma
s
ea
Aq
x
wf
f
~
~~~
d~s~
F
µ
PF
a
S aD d
Ee
P
Fa
s
ea
Aq
x
w f
f
~
~~~
d~ s~
F
m m
PT
1
SaDd
Ee
P
T1
s
ea
Af
f
~
~~~
d~s~F
P
Ma
SaDd
Ee
P
Sp
s
ea
Aq
x
wf
f
~
~~~
d~s~
F
PT
2
S aD d
Ee
P
T2
s
ea
Af
f
~
~~~
d~s~ F
PF
a
S aD d
Ee
P
Sp
s
ea
Aq
x
w f
f
~
~~~
d~ s~
F
mm
P
Ch
S aD d
Ees
AfF
µ µ
mm
m m
P
Ch
SaDd
Ee s
AfF
1/.25
1
Modeling complexity
• The good: Tend to be less biased• The bad: Easy to make scripting or theoretical
mistakes. Are we really modeling what we think we are?? How to know?
![Page 20: Simulating and Modeling Extended Twin Family Data](https://reader031.fdocuments.in/reader031/viewer/2022012916/568144fe550346895db1c95a/html5/thumbnails/20.jpg)
Part II: Simulating complex models
![Page 21: Simulating and Modeling Extended Twin Family Data](https://reader031.fdocuments.in/reader031/viewer/2022012916/568144fe550346895db1c95a/html5/thumbnails/21.jpg)
Independent check of models:Independent check of models: Model validationModel validation: Check that your models work as they are : Check that your models work as they are
supposed to and check the statistical properties of estimatessupposed to and check the statistical properties of estimates Sensitivity analysisSensitivity analysis: Check the effect on parameter estimates : Check the effect on parameter estimates
when assumptions are violated (e.g., different modes of when assumptions are violated (e.g., different modes of assortative mating, genetic action, etc.) assortative mating, genetic action, etc.)
Method for predicting complex dynamics in Method for predicting complex dynamics in population geneticspopulation genetics
Simulation provides knowledge about processes that are difficult/impossible to
figure out analytically
![Page 22: Simulating and Modeling Extended Twin Family Data](https://reader031.fdocuments.in/reader031/viewer/2022012916/568144fe550346895db1c95a/html5/thumbnails/22.jpg)
Simulation program: GeneEvolve
![Page 23: Simulating and Modeling Extended Twin Family Data](https://reader031.fdocuments.in/reader031/viewer/2022012916/568144fe550346895db1c95a/html5/thumbnails/23.jpg)
Implemented in R, open-source, user modifiableImplemented in R, open-source, user modifiable User specifies 31 basic parameters up front (and User specifies 31 basic parameters up front (and
17 advanced ones); no need to alter script after 17 advanced ones); no need to alter script after that.that.
GeneEvolve 0.73
Download: www.matthewckeller.com
![Page 24: Simulating and Modeling Extended Twin Family Data](https://reader031.fdocuments.in/reader031/viewer/2022012916/568144fe550346895db1c95a/html5/thumbnails/24.jpg)
User specifies:User specifies: population size, # generations for population to population size, # generations for population to
evolve, threshold effects, mechanisms of evolve, threshold effects, mechanisms of assortative mating, vertical transmission, etc.assortative mating, vertical transmission, etc.
3 types of genetic effects3 types of genetic effects 5 types of environmental effects5 types of environmental effects 13 types of moderator/covariate effects13 types of moderator/covariate effects
How GeneEvolve works:
![Page 25: Simulating and Modeling Extended Twin Family Data](https://reader031.fdocuments.in/reader031/viewer/2022012916/568144fe550346895db1c95a/html5/thumbnails/25.jpg)
Parameters of interest for present simulationsParameters of interest for present simulations A = additive genetic effectsA = additive genetic effects D = dominance genetic effectsD = dominance genetic effects U = unique environmental effectsU = unique environmental effects F = familial environmental effectsF = familial environmental effects S = sibling environmental effectsS = sibling environmental effects AM = correlation between spousesAM = correlation between spouses am.model = “I”: primary phenotypic am.model = “I”: primary phenotypic
“ “II”: social homogamyII”: social homogamy
How GeneEvolve works:
![Page 26: Simulating and Modeling Extended Twin Family Data](https://reader031.fdocuments.in/reader031/viewer/2022012916/568144fe550346895db1c95a/html5/thumbnails/26.jpg)
At adulthood, ~ x% find mates s.t. correlation b/w At adulthood, ~ x% find mates s.t. correlation b/w mating phenotypes = AM:mating phenotypes = AM:
Pairs have children :Pairs have children : Rate Rate determined by user-specified population growthdetermined by user-specified population growth
Process iterated Process iterated nn times (Markov Chain) times (Markov Chain)
How GeneEvolve works (cont):
![Page 27: Simulating and Modeling Extended Twin Family Data](https://reader031.fdocuments.in/reader031/viewer/2022012916/568144fe550346895db1c95a/html5/thumbnails/27.jpg)
After After n n iterations, population splits into two:iterations, population splits into two: Parents of spousesParents of spouses Parents of twinsParents of twins
Parents of twins have offspring (MZ/DZ twins & Parents of twins have offspring (MZ/DZ twins & their sibs)their sibs)
Twins mate with spousal population & have Twins mate with spousal population & have offspringoffspring
How GeneEvolve works (cont):
![Page 28: Simulating and Modeling Extended Twin Family Data](https://reader031.fdocuments.in/reader031/viewer/2022012916/568144fe550346895db1c95a/html5/thumbnails/28.jpg)
3 generations (grandparents, parents, & offspring) 3 generations (grandparents, parents, & offspring) of phenotypic data written out, one row per family, of phenotypic data written out, one row per family, potentially across repeated measurespotentially across repeated measures
This data can be entered into structural models for This data can be entered into structural models for model validation and sensitivity analysismodel validation and sensitivity analysis
A summary PDF at end showsA summary PDF at end shows: : Basic simulation statisticsBasic simulation statistics Changes in variance components across timeChanges in variance components across time Correlations between 10 relative typesCorrelations between 10 relative types
What you get:
![Page 29: Simulating and Modeling Extended Twin Family Data](https://reader031.fdocuments.in/reader031/viewer/2022012916/568144fe550346895db1c95a/html5/thumbnails/29.jpg)
Many parameters change dynamically Many parameters change dynamically (evolutionarily) as functions of other parameters in (evolutionarily) as functions of other parameters in models that include assortative mating and vertical models that include assortative mating and vertical transmission. Predicting changes in such transmission. Predicting changes in such parameters is impractical and approaches parameters is impractical and approaches impossible in models where many things impossible in models where many things simultaneously going on.simultaneously going on.
Why go through the trouble to simulate a population’s evolution rather than to simulate in one step (or analytically)?
![Page 30: Simulating and Modeling Extended Twin Family Data](https://reader031.fdocuments.in/reader031/viewer/2022012916/568144fe550346895db1c95a/html5/thumbnails/30.jpg)
Getting StartedGetting Started Copy F:/matt/GE folder into your home directory Copy F:/matt/GE folder into your home directory Start R. Then File -> Open Script -> home:GE/GE-73.RStart R. Then File -> Open Script -> home:GE/GE-73.R
Running GeneEvolveRunning GeneEvolve
1.1. Create a reality where A=.3, D=.2, F=.1, S=.1, U=.3 & Create a reality where A=.3, D=.2, F=.1, S=.1, U=.3 & AM = .2 (all other parameters = 0). After it runs (~1.5 AM = .2 (all other parameters = 0). After it runs (~1.5 min), open the resulting PDF. What happens to the A min), open the resulting PDF. What happens to the A variation across 10 generations? D? F? S? Why?variation across 10 generations? D? F? S? Why?
2.2. Do the same thing but change the mode of mating to Do the same thing but change the mode of mating to social homogamy (am.model <- “II”). What happens?social homogamy (am.model <- “II”). What happens?
3.3. Run another model you find interesting & see what Run another model you find interesting & see what happenshappens
GeneEvolve Practical
![Page 31: Simulating and Modeling Extended Twin Family Data](https://reader031.fdocuments.in/reader031/viewer/2022012916/568144fe550346895db1c95a/html5/thumbnails/31.jpg)
Why does variance of A increase in presence of AM?
![Page 32: Simulating and Modeling Extended Twin Family Data](https://reader031.fdocuments.in/reader031/viewer/2022012916/568144fe550346895db1c95a/html5/thumbnails/32.jpg)
Part III: Model validation and sensitivity analysis of extended twin family models
![Page 33: Simulating and Modeling Extended Twin Family Data](https://reader031.fdocuments.in/reader031/viewer/2022012916/568144fe550346895db1c95a/html5/thumbnails/33.jpg)
Using complex models without independent validation (e.g., simulation) is like…
QuickTime™ and a decompressor
are needed to see this picture.
![Page 34: Simulating and Modeling Extended Twin Family Data](https://reader031.fdocuments.in/reader031/viewer/2022012916/568144fe550346895db1c95a/html5/thumbnails/34.jpg)
Process of model validation
1.1. Simulate a dataset that has parameters that your Simulate a dataset that has parameters that your model can estimate.model can estimate.
2.2. Run your model on the simulated datasetRun your model on the simulated dataset
3.3. Obtain and store parameter estimatesObtain and store parameter estimates
4.4. Repeat steps 1-3 many (e.g., 1000) times Repeat steps 1-3 many (e.g., 1000) times
![Page 35: Simulating and Modeling Extended Twin Family Data](https://reader031.fdocuments.in/reader031/viewer/2022012916/568144fe550346895db1c95a/html5/thumbnails/35.jpg)
Results of model validation
If the mean parameter estimate = the simulated If the mean parameter estimate = the simulated parameter estimate, the estimate is parameter estimate, the estimate is unbiased. unbiased. If If your model has no mistakes, parameters should your model has no mistakes, parameters should generally be unbiased (there are exceptions)generally be unbiased (there are exceptions)
The standard deviation of an estimates The standard deviation of an estimates corresponds to its corresponds to its standard errorstandard error and its and its distribution to its distribution to its sampling distributionsampling distribution
You can also easily study the You can also easily study the multivariate multivariate sampling distribution and statisticssampling distribution and statistics. E.g., how . E.g., how correlated parameters are.correlated parameters are.
![Page 36: Simulating and Modeling Extended Twin Family Data](https://reader031.fdocuments.in/reader031/viewer/2022012916/568144fe550346895db1c95a/html5/thumbnails/36.jpg)
Graphical representation: model validation
Simulate
parameters & get
simulated dataset
A S U
.5
.4
.3
.2
.1
.0
![Page 37: Simulating and Modeling Extended Twin Family Data](https://reader031.fdocuments.in/reader031/viewer/2022012916/568144fe550346895db1c95a/html5/thumbnails/37.jpg)
Graphical representation: model validation
Simulate
parameters & get
simulated dataset
Run Mx, get
estimates
A D U
.5
.4
.3
.2
.1
.0
e.g., A=.34, D=.17,U=.49
![Page 38: Simulating and Modeling Extended Twin Family Data](https://reader031.fdocuments.in/reader031/viewer/2022012916/568144fe550346895db1c95a/html5/thumbnails/38.jpg)
Graphical representation: model validation
Simulate
parameters & get
simulated dataset
Run Mx, get
estimates
A D U
.5
.4
.3
.2
.1
.0
e.g., A=.34, D=.17,U=.49
Repeat 1
e.g., A=.31, D=.19,U=.50
![Page 39: Simulating and Modeling Extended Twin Family Data](https://reader031.fdocuments.in/reader031/viewer/2022012916/568144fe550346895db1c95a/html5/thumbnails/39.jpg)
Graphical representation: model validation
Simulate
parameters & get
simulated dataset
Run Mx, get
estimates
A D U
.5
.4
.3
.2
.1
.0
e.g., A=.34, D=.17,U=.49
Repeat 1
Repeat 2
e.g., A=.31, D=.19,U=.50e.g., A=.25, D=.26,U=.48
![Page 40: Simulating and Modeling Extended Twin Family Data](https://reader031.fdocuments.in/reader031/viewer/2022012916/568144fe550346895db1c95a/html5/thumbnails/40.jpg)
Graphical representation: model validation
Simulate
parameters & get
simulated dataset
Run Mx, get
estimates
A D U
.5
.4
.3
.2
.1
.0
e.g., A=.34, D=.17,U=.49
Repeat 1
Repeat 2
Repeat 3e.g., A=.31, D=.19,U=.50e.g., A=.25, D=.26,U=.48e.g., A=.29, D=.21,U=.51
![Page 41: Simulating and Modeling Extended Twin Family Data](https://reader031.fdocuments.in/reader031/viewer/2022012916/568144fe550346895db1c95a/html5/thumbnails/41.jpg)
Graphical representation: model validation
Simulate
parameters & get
simulated dataset
Run Mx, get
estimates
A D U
.5
.4
.3
.2
.1
.0
Repeat 1
Repeat 2
Repeat 3
.
.
.
Repeat 20
![Page 42: Simulating and Modeling Extended Twin Family Data](https://reader031.fdocuments.in/reader031/viewer/2022012916/568144fe550346895db1c95a/html5/thumbnails/42.jpg)
Graphical representation: model validation
Simulate
parameters & get
simulated dataset
Run Mx, get
estimates
A D U
.5
.4
.3
.2
.1
.0
Repeat 1
Repeat 2
Repeat 3
.
.
.
Repeat 1000
![Page 43: Simulating and Modeling Extended Twin Family Data](https://reader031.fdocuments.in/reader031/viewer/2022012916/568144fe550346895db1c95a/html5/thumbnails/43.jpg)
Graphical representation: model validation
Simulate
parameters & get
simulated dataset
Run Mx, get
estimates
A D U
.5
.4
.3
.2
.1
.0
Repeat 1
Repeat 2
Repeat 3
.
.
.
Repeat 1000
Mean estimate
~ true estimate:
Unbiased!
![Page 44: Simulating and Modeling Extended Twin Family Data](https://reader031.fdocuments.in/reader031/viewer/2022012916/568144fe550346895db1c95a/html5/thumbnails/44.jpg)
Graphical representation: model validation
Simulate
parameters & get
simulated dataset
Run Mx, get
estimates
A D U
.5
.4
.3
.2
.1
.0
Repeat 1
Repeat 2
Repeat 3
.
.
.
Repeat 1000
Boxplots give
idea about
variance and
shape of
sampling
distributions
![Page 45: Simulating and Modeling Extended Twin Family Data](https://reader031.fdocuments.in/reader031/viewer/2022012916/568144fe550346895db1c95a/html5/thumbnails/45.jpg)
Graphical representation: model validation & multivariate distributions
Simulate
parameters & get
simulated dataset
A
D
Simulate
parameters & get
simulated dataset
Run Mx, get
estimatese.g., A=.34, D=.17,U=.49
![Page 46: Simulating and Modeling Extended Twin Family Data](https://reader031.fdocuments.in/reader031/viewer/2022012916/568144fe550346895db1c95a/html5/thumbnails/46.jpg)
Graphical representation: model validation & multivariate distributions
Simulate
parameters & get
simulated dataset
A
D
Simulate
parameters & get
simulated dataset
Run Mx, get
estimatese.g., A=.34, D=.17,U=.49
Repeat 1
e.g., A=.31, D=.19,U=.50
![Page 47: Simulating and Modeling Extended Twin Family Data](https://reader031.fdocuments.in/reader031/viewer/2022012916/568144fe550346895db1c95a/html5/thumbnails/47.jpg)
Graphical representation: model validation & multivariate distributions
Simulate
parameters & get
simulated dataset
A
D
Simulate
parameters & get
simulated dataset
Run Mx, get
estimates
Repeat 1
Repeat 2
Repeat 3
.
.
.
Repeat 20
![Page 48: Simulating and Modeling Extended Twin Family Data](https://reader031.fdocuments.in/reader031/viewer/2022012916/568144fe550346895db1c95a/html5/thumbnails/48.jpg)
Graphical representation: model validation & multivariate distributions
Simulate
parameters & get
simulated dataset
A
D
Simulate
parameters & get
simulated dataset
Run Mx, get
estimates
Repeat 1
Repeat 2
Repeat 3
.
.
.
Repeat 1000
A & D estimates
negatively correlated -
suggesting they use
overlapping information to
be estimated
![Page 49: Simulating and Modeling Extended Twin Family Data](https://reader031.fdocuments.in/reader031/viewer/2022012916/568144fe550346895db1c95a/html5/thumbnails/49.jpg)
Process of sensitivity analysis
1.1. Simulate a dataset that has one or more parameters Simulate a dataset that has one or more parameters that your model that your model cannotcannot estimate. estimate.
2.2. Run your model on the simulated datasetRun your model on the simulated dataset
3.3. Obtain and store parameter estimatesObtain and store parameter estimates
4.4. Repeat steps 1-3 many (e.g., 1000) times Repeat steps 1-3 many (e.g., 1000) times
![Page 50: Simulating and Modeling Extended Twin Family Data](https://reader031.fdocuments.in/reader031/viewer/2022012916/568144fe550346895db1c95a/html5/thumbnails/50.jpg)
Results of sensitivity analysis
Because we are simulating Because we are simulating violations of violations of assumptionsassumptions, we expect parameters to be biased, we expect parameters to be biased . . The question becomes: The question becomes: howhow biased? I.e., how big biased? I.e., how big of a deal are these violations? We should be able of a deal are these violations? We should be able to quantify the answers to these questions.to quantify the answers to these questions.
![Page 51: Simulating and Modeling Extended Twin Family Data](https://reader031.fdocuments.in/reader031/viewer/2022012916/568144fe550346895db1c95a/html5/thumbnails/51.jpg)
Graphical representation: model sensitivitySimulate
parameters that
include a violation
(here, both D & S
exist
simultaneously) &
get simulated
dataset
Run Mx, get
estimates
A D S U
.5
.4
.3
.2
.1
.0
Repeat 1
Repeat 2
Repeat 3
.
.
.
Repeat 1000
![Page 52: Simulating and Modeling Extended Twin Family Data](https://reader031.fdocuments.in/reader031/viewer/2022012916/568144fe550346895db1c95a/html5/thumbnails/52.jpg)
Run GeneEvolveRun GeneEvolve
1.1. Create a reality where A=.4, D=.1, S=.2, U=.3 & AM = 0. Datasets Create a reality where A=.4, D=.1, S=.2, U=.3 & AM = 0. Datasets MZM, MZF, DZM, DZF, DZOS are made automatically.MZM, MZF, DZM, DZF, DZOS are made automatically.
Run MxRun Mx
1.1. Run the script “GE.Twin_ASE.mx”. This is an ASE script where D Run the script “GE.Twin_ASE.mx”. This is an ASE script where D is fixed to 0is fixed to 0
2.2. Run the script “NTF.mx” This is a nuclear twin family script where Run the script “NTF.mx” This is a nuclear twin family script where A, D, and S are simultaneously estimated. A, D, and S are simultaneously estimated.
3.3. Once you have estimates of A, D, and S from both scripts, come up Once you have estimates of A, D, and S from both scripts, come up and write them into the Excel spreadsheet. They are found in the and write them into the Excel spreadsheet. They are found in the 7th, 8th, and 9th elements of the P matrix in “GE.Twin_ASE.mx” 7th, 8th, and 9th elements of the P matrix in “GE.Twin_ASE.mx” and ??? In the ??? matrix in “NTF.mx” and ??? In the ??? matrix in “NTF.mx”
Sensitivity analysis practical
![Page 53: Simulating and Modeling Extended Twin Family Data](https://reader031.fdocuments.in/reader031/viewer/2022012916/568144fe550346895db1c95a/html5/thumbnails/53.jpg)
If there’s time…Model validation and sensitivity
results for 4 models
![Page 54: Simulating and Modeling Extended Twin Family Data](https://reader031.fdocuments.in/reader031/viewer/2022012916/568144fe550346895db1c95a/html5/thumbnails/54.jpg)
Reality: A=.5, D=.2
![Page 55: Simulating and Modeling Extended Twin Family Data](https://reader031.fdocuments.in/reader031/viewer/2022012916/568144fe550346895db1c95a/html5/thumbnails/55.jpg)
Reality: A=.5, S=.2
![Page 56: Simulating and Modeling Extended Twin Family Data](https://reader031.fdocuments.in/reader031/viewer/2022012916/568144fe550346895db1c95a/html5/thumbnails/56.jpg)
Reality: A=.4, D=.15, S=.15
![Page 57: Simulating and Modeling Extended Twin Family Data](https://reader031.fdocuments.in/reader031/viewer/2022012916/568144fe550346895db1c95a/html5/thumbnails/57.jpg)
Reality: A=.35, D=.15, F=.2, S=.15, T=.15, AM=.3
![Page 58: Simulating and Modeling Extended Twin Family Data](https://reader031.fdocuments.in/reader031/viewer/2022012916/568144fe550346895db1c95a/html5/thumbnails/58.jpg)
Reality: A=.45, D=.15, F=.25, AM=.3 (Soc Hom)
![Page 59: Simulating and Modeling Extended Twin Family Data](https://reader031.fdocuments.in/reader031/viewer/2022012916/568144fe550346895db1c95a/html5/thumbnails/59.jpg)
Reality: A=.4, A*A=.15, S=.15
![Page 60: Simulating and Modeling Extended Twin Family Data](https://reader031.fdocuments.in/reader031/viewer/2022012916/568144fe550346895db1c95a/html5/thumbnails/60.jpg)
Reality: A=.4, A*Age=.15, S=.15
![Page 61: Simulating and Modeling Extended Twin Family Data](https://reader031.fdocuments.in/reader031/viewer/2022012916/568144fe550346895db1c95a/html5/thumbnails/61.jpg)
A,D, & F estimates are highly correlated in Stealth & Cascade models
![Page 62: Simulating and Modeling Extended Twin Family Data](https://reader031.fdocuments.in/reader031/viewer/2022012916/568144fe550346895db1c95a/html5/thumbnails/62.jpg)
Simulation is not a panacea Simulation can be said to provide “knowledge Simulation can be said to provide “knowledge
without understanding.” It is a helpful tool for without understanding.” It is a helpful tool for understanding, but doesn’t provide understanding understanding, but doesn’t provide understanding in and of itself.in and of itself.
Simulations themselves rely on assumptions about Simulations themselves rely on assumptions about how processes work. If these are wrong, our how processes work. If these are wrong, our simulation results may not reflect reality. simulation results may not reflect reality.
![Page 63: Simulating and Modeling Extended Twin Family Data](https://reader031.fdocuments.in/reader031/viewer/2022012916/568144fe550346895db1c95a/html5/thumbnails/63.jpg)
Sex limitation possible only for A at the momentSex limitation possible only for A at the moment No multivariate except for longitudinalNo multivariate except for longitudinal
GeneEvolve Limitations
Download: www.matthewckeller.com
![Page 64: Simulating and Modeling Extended Twin Family Data](https://reader031.fdocuments.in/reader031/viewer/2022012916/568144fe550346895db1c95a/html5/thumbnails/64.jpg)
All models require assumptions. Generally, more All models require assumptions. Generally, more assumptions = more biased estimatesassumptions = more biased estimates
Extended twin family designs require fewer Extended twin family designs require fewer assumptions and tend to be less biasedassumptions and tend to be less biased
Simulation is a powerful tool for checking comlex Simulation is a powerful tool for checking comlex models (and not just extended twin family models)models (and not just extended twin family models)
Conclusions
![Page 65: Simulating and Modeling Extended Twin Family Data](https://reader031.fdocuments.in/reader031/viewer/2022012916/568144fe550346895db1c95a/html5/thumbnails/65.jpg)
Why does variance of A increase in presence of AM?
PMa
Fa
Dd
Ee f
A
qw
PFa
Fa
D d
Eef
A
qw
m
m
PT1
Fa
Dd
Ee f
A
PT2
Fa
Dd
Eef
A
mm
1/.25
µ
Answer: When two spouses are
phenotypically similar, they also tend
to have similar A effects.
Offspring A is a weighted sum of
parental A.
Therefore, variance of A increases for
same reason that the variance of any
sum increases when components are
correlated.
For similar reasons, variance of the
other transmitted parameter, F, also
increases.