setwd("~/Documents/Teresa/MedScholars/Data/Data files/CSV")
srlm1 <- read.csv("working data_SRLM_APOE clean.csv")
srlm2 <- read.csv("working data_SRLM_ms_age.csv")Predicting SRLM_avg:
Gene x Gender ANOVA
s2 <- aov(SRLM_avg ~ sex + ApoE + sex:ApoE, data = srlm1)
summary (s2)## Df Sum Sq Mean Sq F value Pr(>F)
## sex 1 0.0085 0.008526 1.432 0.2346
## ApoE 3 0.0511 0.017027 2.861 0.0415 *
## sex:ApoE 3 0.0103 0.003448 0.579 0.6302
## Residuals 87 0.5179 0.005952
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1Gender x Dx ANOVA
s3 <- aov(SRLM_avg ~ sex + Dx + Dx:sex, data = srlm1)
summary (s3)## Df Sum Sq Mean Sq F value Pr(>F)
## sex 1 0.0085 0.00853 1.897 0.172
## Dx 2 0.1774 0.08869 19.736 8.05e-08 ***
## sex:Dx 2 0.0020 0.00099 0.220 0.803
## Residuals 89 0.3999 0.00449
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1Gender x Gene x Dx ANOVA
s5 <- aov(SRLM_avg ~(Dx*sex*ApoE), data = srlm1)
summary (s5)## Df Sum Sq Mean Sq F value Pr(>F)
## Dx 2 0.1742 0.08708 19.191 2.09e-07 ***
## sex 1 0.0117 0.01173 2.586 0.112
## ApoE 3 0.0081 0.00269 0.593 0.621
## Dx:sex 2 0.0021 0.00104 0.229 0.796
## Dx:ApoE 6 0.0188 0.00313 0.691 0.658
## sex:ApoE 3 0.0073 0.00244 0.539 0.657
## Dx:sex:ApoE 5 0.0389 0.00778 1.714 0.142
## Residuals 72 0.3267 0.00454
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1Linear Regression with Dx, Age, and Education I:
s4 <- lm (SRLM_avg ~ Dx + Age + education, data = srlm1)
summary (s4)##
## Call:
## lm(formula = SRLM_avg ~ Dx + Age + education, data = srlm1)
##
## Residuals:
## Min 1Q Median 3Q Max
## -0.112238 -0.050112 -0.003745 0.025627 0.171011
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 0.5554407 0.0746436 7.441 5.99e-11 ***
## DxMCI 0.0362519 0.0191302 1.895 0.0613 .
## DxNC 0.0993279 0.0176615 5.624 2.13e-07 ***
## Age -0.0013144 0.0008574 -1.533 0.1288
## education 0.0006396 0.0033800 0.189 0.8503
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 0.06691 on 89 degrees of freedom
## (1 observation deleted due to missingness)
## Multiple R-squared: 0.3123, Adjusted R-squared: 0.2814
## F-statistic: 10.1 on 4 and 89 DF, p-value: 8.652e-07II: Age is significant in predicting RIGHT, but not LEFT SRLM (not ERC) atrophy, independent of education and dx
s14 <- lm(SRLM_r~Age+education+Dx, data = srlm1)
summary (s14)##
## Call:
## lm(formula = SRLM_r ~ Age + education + Dx, data = srlm1)
##
## Residuals:
## Min 1Q Median 3Q Max
## -0.117835 -0.048941 -0.007725 0.041137 0.244788
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 0.6101279 0.0808992 7.542 3.75e-11 ***
## Age -0.0020401 0.0009292 -2.196 0.0307 *
## education 0.0006533 0.0036633 0.178 0.8589
## DxMCI 0.0295130 0.0207335 1.423 0.1581
## DxNC 0.0802550 0.0191417 4.193 6.49e-05 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 0.07251 on 89 degrees of freedom
## (1 observation deleted due to missingness)
## Multiple R-squared: 0.2331, Adjusted R-squared: 0.1986
## F-statistic: 6.762 on 4 and 89 DF, p-value: 8.463e-05s15 <-lm(SRLM_l~Age+education+Dx, data = srlm1)
summary (s15)##
## Call:
## lm(formula = SRLM_l ~ Age + education + Dx, data = srlm1)
##
## Residuals:
## Min 1Q Median 3Q Max
## -0.124077 -0.058623 -0.001447 0.037910 0.270176
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 0.5007534 0.0887730 5.641 1.98e-07 ***
## Age -0.0005887 0.0010197 -0.577 0.5651
## education 0.0006259 0.0040198 0.156 0.8766
## DxMCI 0.0429907 0.0227514 1.890 0.0621 .
## DxNC 0.1184008 0.0210047 5.637 2.01e-07 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 0.07957 on 89 degrees of freedom
## (1 observation deleted due to missingness)
## Multiple R-squared: 0.2957, Adjusted R-squared: 0.264
## F-statistic: 9.34 on 4 and 89 DF, p-value: 2.383e-06Median split on Age (now age is significant for srlm_avg)
s7 <- aov(SRLM_avg ~ sex + Age + Age: sex, data = srlm2)
summary (s7)## Df Sum Sq Mean Sq F value Pr(>F)
## sex 1 0.0078 0.00779 1.301 0.2571
## Age 1 0.0379 0.03794 6.333 0.0136 *
## sex:Age 1 0.0007 0.00066 0.110 0.7406
## Residuals 90 0.5392 0.00599
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1s8 <- aov(SRLM_avg ~ Age + Dx + Age:Dx, data = srlm2)
summary (s8)## Df Sum Sq Mean Sq F value Pr(>F)
## Age 1 0.0400 0.04001 9.096 0.00335 **
## Dx 2 0.1554 0.07772 17.669 3.54e-07 ***
## Age:Dx 2 0.0030 0.00150 0.342 0.71157
## Residuals 88 0.3871 0.00440
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1Predicting ERC (slight trend in Dx x Gene interaction)
s9<- aov(ERC_avg ~ (Dx*sex*ApoE), data=srlm1)
summary (s9)## Df Sum Sq Mean Sq F value Pr(>F)
## Dx 2 0.891 0.4453 5.453 0.0063 **
## sex 1 0.084 0.0836 1.024 0.3151
## ApoE 3 0.181 0.0603 0.738 0.5329
## Dx:sex 2 0.194 0.0969 1.187 0.3112
## Dx:ApoE 6 0.955 0.1591 1.948 0.0850 .
## sex:ApoE 3 0.347 0.1158 1.418 0.2449
## Dx:sex:ApoE 4 0.594 0.1484 1.817 0.1353
## Residuals 70 5.717 0.0817
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 3 observations deleted due to missingnessPredicting average hippocampal volume (trend in main effect of Gender, and trend in Dx x APOE)
s10 <- aov(Hipp_avg~(Dx*sex*ApoE), data = srlm1)
summary (s10)## Df Sum Sq Mean Sq F value Pr(>F)
## Dx 2 0.1667 0.08337 13.142 1.43e-05 ***
## sex 1 0.0212 0.02122 3.345 0.0717 .
## ApoE 3 0.0671 0.02238 3.527 0.0192 *
## Dx:sex 2 0.0048 0.00242 0.382 0.6840
## Dx:ApoE 6 0.0796 0.01327 2.091 0.0651 .
## sex:ApoE 3 0.0209 0.00697 1.099 0.3554
## Dx:sex:ApoE 4 0.0395 0.00987 1.556 0.1957
## Residuals 70 0.4441 0.00634
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 3 observations deleted due to missingnessMain effect of gender on right hippocampal volume = trend. But no main effect of gender on right or left SRLM.
s11 <- aov(Hipp_r~sex, data = srlm2)
summary (s11)## Df Sum Sq Mean Sq F value Pr(>F)
## sex 1 0.0309 0.03094 3.086 0.0824 .
## Residuals 89 0.8923 0.01002
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 3 observations deleted due to missingnesss13 <- aov(SRLM_r~(sex*ApoE), data = srlm1)
summary (s13)## Df Sum Sq Mean Sq F value Pr(>F)
## sex 1 0.0078 0.007778 1.156 0.285
## ApoE 3 0.0233 0.007762 1.154 0.332
## sex:ApoE 3 0.0207 0.006887 1.024 0.386
## Residuals 87 0.5854 0.006728s13.1 <- aov(SRLM_l~(sex*ApoE), data = srlm1)
summary (s13.1)## Df Sum Sq Mean Sq F value Pr(>F)
## sex 1 0.0093 0.00931 1.177 0.28102
## ApoE 3 0.0957 0.03192 4.035 0.00976 **
## sex:ApoE 3 0.0073 0.00242 0.306 0.82098
## Residuals 87 0.6881 0.00791
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1But a trend in Dx x Gender x APOE is observed in SRLM_R.
s16 <- aov(SRLM_r~(Dx*sex*ApoE), data=srlm1)
summary (s16)## Df Sum Sq Mean Sq F value Pr(>F)
## Dx 2 0.1234 0.06168 11.091 6.33e-05 ***
## sex 1 0.0108 0.01083 1.948 0.1671
## ApoE 3 0.0053 0.00178 0.320 0.8109
## Dx:sex 2 0.0032 0.00158 0.285 0.7531
## Dx:ApoE 6 0.0264 0.00440 0.792 0.5791
## sex:ApoE 3 0.0099 0.00330 0.594 0.6210
## Dx:sex:ApoE 5 0.0577 0.01153 2.074 0.0786 .
## Residuals 72 0.4004 0.00556
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1However, there is no interaction between age and gender on SRLM_R thinning (with median split on age)
s17 <- aov(SRLM_r~(sex*Age), data=srlm2)
summary (s17)## Df Sum Sq Mean Sq F value Pr(>F)
## sex 1 0.0078 0.00784 1.255 0.26561
## Age 1 0.0627 0.06272 10.040 0.00209 **
## sex:Age 1 0.0043 0.00429 0.687 0.40954
## Residuals 90 0.5622 0.00625
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1Note that the echo = FALSE parameter was added to the code chunk to prevent printing of the R code that generated the plot.