Download presentation
Presentation is loading. Please wait.
1
Genetic simplex model: practical
Sanja Franic, Conor Dolan
2
Practical: estimate the genetic and environmental contributions to temporal stability and change in full-scale IQ measured at four time points (mean ages 5.5, 6.8, 9.7, 12.2, SDs .3, .19, .43, .24) N = 562 twin pairs (261 MZ, 301 DZ) The proportions of observed FSIQ data: 0.812, 0.295, 0.490, (MZ twin 1) 0.812, 0.295, 0.490, (MZ twin 2) 0.774, 0.379, 0.598, (DZ twin1) 0.774, 0.379, 0.598, (DZ twin 2)
3
Models Saturated Model
- estimate means (subject to twin1-twin2 equality constraints), variances and covariances IQ11 IQ21 IQ31 IQ41 IQ12 IQ22 IQ32 IQ42 var1 cov21 var2 cov31 cov32 var3 cov41 cov42 cov43 var4 cov51 cov52 cov53 cov54 var5 cov61 cov62 cov63 cov64 cov65 var6 IQ23 cov71 cov72 cov73 cov74 cov75 cov76 var7 IQ24 cov81 cov82 cov83 cov84 cov85 cov86 cov87 var8 different over MZ & DZ IQ11 IQ21 IQ31 IQ41 IQ12 IQ22 IQ32 IQ42 mean m1 m2 m3 m4
4
Models Saturated Model
- estimate means (subject to twin1-twin1 equality constraints), variances and covariances IQ11 IQ21 IQ31 IQ41 IQ12 IQ22 IQ32 IQ42 var1 cov21 var2 cov31 cov32 var3 cov41 cov42 cov43 var4 cov51 cov52 cov53 cov54 var5 cov61 cov62 cov63 cov64 cov65 var6 IQ23 cov71 cov72 cov73 cov74 cov75 cov76 var7 IQ24 cov81 cov82 cov83 cov84 cov85 cov86 cov87 var8 different over MZ & DZ IQ11 IQ21 IQ31 IQ41 IQ12 IQ22 IQ32 IQ42 mean m1 m2 m3 m4 How many parameters?
5
Models Saturated Model
- estimate means (subject to twin1-twin1 equality constraints), variances and covariances IQ11 IQ21 IQ31 IQ41 IQ12 IQ22 IQ32 IQ42 var1 cov21 var2 cov31 cov32 var3 cov41 cov42 cov43 var4 cov51 cov52 cov53 cov54 var5 cov61 cov62 cov63 cov64 cov65 var6 IQ23 cov71 cov72 cov73 cov74 cov75 cov76 var7 IQ24 cov81 cov82 cov83 cov84 cov85 cov86 cov87 var8 ( 8*(8-1)/2 + 8 ) * 2 different over MZ & DZ IQ11 IQ21 IQ31 IQ41 IQ12 IQ22 IQ32 IQ42 mean m1 m2 m3 m4 How many parameters?
6
Models Saturated Model
- estimate means (subject to twin1-twin1 equality constraints), variances and covariances IQ11 IQ21 IQ31 IQ41 IQ12 IQ22 IQ32 IQ42 var1 cov21 var2 cov31 cov32 var3 cov41 cov42 cov43 var4 cov51 cov52 cov53 cov54 var5 cov61 cov62 cov63 cov64 cov65 var6 IQ23 cov71 cov72 cov73 cov74 cov75 cov76 var7 IQ24 cov81 cov82 cov83 cov84 cov85 cov86 cov87 var8 ( 8*(8-1)/2 + 8 ) * 2 different over MZ & DZ IQ11 IQ21 IQ31 IQ41 IQ12 IQ22 IQ32 IQ42 mean m1 m2 m3 m4 How many parameters? 4
7
Models Saturated Model
- estimate means (subject to twin1-twin1 equality constraints), variances and covariances IQ11 IQ21 IQ31 IQ41 IQ12 IQ22 IQ32 IQ42 var1 cov21 var2 cov31 cov32 var3 cov41 cov42 cov43 var4 cov51 cov52 cov53 cov54 var5 cov61 cov62 cov63 cov64 cov65 var6 IQ23 cov71 cov72 cov73 cov74 cov75 cov76 var7 IQ24 cov81 cov82 cov83 cov84 cov85 cov86 cov87 var8 ( 8*(8-1)/2 + 8 ) * 2 different over MZ & DZ IQ11 IQ21 IQ31 IQ41 IQ12 IQ22 IQ32 IQ42 mean m1 m2 m3 m4 How many parameters? 76 4
8
Models Saturated Model
- estimate means (subject to twin1-twin1 equality constraints), variances and covariances IQ11 IQ21 IQ31 IQ41 IQ12 IQ22 IQ32 IQ42 var1 cov21 var2 cov31 cov32 var3 cov41 cov42 cov43 var4 cov51 cov52 cov53 cov54 var5 cov61 cov62 cov63 cov64 cov65 var6 IQ23 cov71 cov72 cov73 cov74 cov75 cov76 var7 IQ24 cov81 cov82 cov83 cov84 cov85 cov86 cov87 var8 ( 8*(8-1)/2 + 8 ) * 2 different over MZ & DZ IQ11 IQ21 IQ31 IQ41 IQ12 IQ22 IQ32 IQ42 mean m1 m2 m3 m4 How many parameters? 76 4
9
Models SMZ = SA+SC+SE SA+SC SA+SC SA+SC+SE SDZ = SA+SC+SE .5SA+SC
2) ACE Cholesky Model - depicting only A component to avoid clutter - means subject to same equality constraints as in the saturated model SMZ = SA+SC+SE SA+SC SA+SC SA+SC+SE SDZ = SA+SC+SE .5SA+SC .5SA+SC SA+SC+SE
10
Models SA = DADAt DA = d11 0 0 0 d21 d22 0 0 d31 d32 d33 0
2) ACE Cholesky Model - depicting only A component to avoid clutter - means subject to same equality constraints as in the saturated model SA = DADAt DA = d d21 d d31 d32 d33 0 d41 d42 d43 d44
11
Models SA = DADAt DA = d11 0 0 0 d21 d22 0 0 d31 d32 d33 0
2) ACE Cholesky Model - depicting only A component to avoid clutter - means subject to same equality constraints as in the saturated model SA = DADAt DA = d d21 d d31 d32 d33 0 d41 d42 d43 d44 How many parameters?
12
Models SA = DADAt DA = d11 0 0 0 d21 d22 0 0 d31 d32 d33 0
2) ACE Cholesky Model - depicting only A component to avoid clutter - means subject to same equality constraints as in the saturated model SA = DADAt DA = d d21 d d31 d32 d33 0 d41 d42 d43 d44 How many parameters? 34
13
Models SMZ = SA+SC+SE SA+SC SA+SC SA+SC+SE SDZ = SA+SC+SE .5SA+SC
3) Simplex Model - the genetic A simplex: 3 + 4 = 7 parameters SMZ = SA+SC+SE SA+SC SA+SC SA+SC+SE SDZ = SA+SC+SE .5SA+SC .5SA+SC SA+SC+SE SA = (I-BA) YA (I-BA) t + QA
14
Models SA = (I-BA) YA (I-BA) t + QA 3) Simplex Model s2z1 s2z2 s2z3
s2z2 s2z3 s2z4 bA21 bA32 bA43 s2a1 s2a2 s2a3 s2a4 SA = (I-BA) YA (I-BA) t + QA
15
Models SMZ = SA+SC+SE SA+SC SA+SC SA+SC+SE SDZ = SA+SC+SE .5SA+SC
3) Simplex Model - the genetic C simplex: 3 + 4 = 7 parameters SMZ = SA+SC+SE SA+SC SA+SC SA+SC+SE SDZ = SA+SC+SE .5SA+SC .5SA+SC SA+SC+SE SC = (I-BC) YC (I-BC) t + QC
16
Models SMZ = SA+SC+SE SA+SC SA+SC SA+SC+SE SDZ = SA+SC+SE .5SA+SC
3) Simplex Model - the genetic E model: 4 parameters SMZ = SA+SC+SE SA+SC SA+SC SA+SC+SE SDZ = SA+SC+SE .5SA+SC .5SA+SC SA+SC+SE SE = QE
17
faculty/sanja/2016/Simplex/Practical/simplexPractical.R
18
Saturated model: Rmz [1,] [2,] [3,] [4,] [5,] [6,] [7,] [8,] Rdz [1,] [2,] [3,] [4,] [5,] [6,] [7,] [8,] 5.5y 6.8y 9.7y 12.2y MZ FSIQ correlation (FIML estimates) DZ FSIQ correlation (FIML estimates)
![Saturated model: Rmz. [1,] [2,]](http://slideplayer.com./slide/12743673/77/images/18/Saturated+model%3A+Rmz.+%5B1%2C%5D+%5B2%2C%5D.jpg "[3,] [4,] [5,] [6,] [7,] [8,] Rdz. [1,] [2,] [3,] [4,] [5,] [6,] [7,] [8,] y 6.8y 9.7y 12.2y MZ FSIQ correlation (FIML estimates) DZ FSIQ correlation (FIML estimates)")
19
ACE Cholesky model: RA_est 5.5y 6.8y 9.7y y RC_est 5.5y 6.8y 9.7y y RE_est 5.5y 6.8y 9.7y y
20
SA = (I-BA) YA (I-BA) t + QA
Simplex model: YA [1,] [2,] [3,] [4,] BA [1,] [2,] [3,] [4,] QA (fixed) [1,] [2,] [3,] [4,] SA = (I-BA) YA (I-BA) t + QA
21
SC = (I-BC) YC (I-BC) t + QC
Simplex model: YC [1,] [2,] [3,] [4,] BC [1,] [2,] [3,] [4,] QC (fixed) [1,] [2,] [3,] [4,] SC = (I-BC) YC (I-BC) t + QC
22
SE = QE Simplex model: YE (fixed) [1,] 0 0 0 0 [2,] 0 0 0 0
[1,] [2,] [3,] [4,] BE (fixed) QE [1,] [2,] [3,] [4,] SE = QE
![SE = QE Simplex model: YE (fixed) [1,] [2,]](http://slideplayer.com./slide/12743673/77/images/22/SE+%3D+QE+Simplex+model%3A+YE+%28fixed%29+%5B1%2C%5D+%5B2%2C%5D.jpg "[1,] [2,] [3,] [4,] BE (fixed) QE. [1,] [2,] [3,] [4,] SE = QE.")
23
Simplex model: Standardized variance components 5.5y 6.8y 9.7y 12.2y h Increasing c Decreasing e About constant
24
Simplex model: Phenotypic covariance between age 9.7 and 12.2 equals Contributions of A to is 77.4% Contributions of A to is 22.6% Contributions of E to is 0% Conclude? > round(SA_est/Sph_est,3) [1,] [2,] [3,] [4,] > round(SC_est/Sph_est,3) [1,] [2,] [3,] [4,] > round(SE_est/Sph_est,3) [1,] [2,] [3,] [4,] Standardized variance components 5.5y 6.8y 9.7y 12.2y h Increasing c Decreasing e About constant
Similar presentations
