-
Notifications
You must be signed in to change notification settings - Fork 0
/
fig6.R
135 lines (109 loc) · 5.06 KB
/
fig6.R
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
rm(list=ls())
################################
# Load workspace and libraries #
################################
load('data/0_finalData.RData')
library(ggplot2)
library(ppcor)
library(mgcv)
library(visreg)
library(scatterplot3d)
library(scales)
####################################
# Lausanne atlas for HIGH fidelity #
####################################
# all individual data points
for (j in c(83,129,234,463,1015))
{
for (i in c(999,9,5))
{
filepath <- paste0('/data/Lausanne',j,'_resource_efficiency0',i,'.txt')
degreePath <- paste0('/data/degreeLausanne',j,'.txt')
degree <- read.csv(degreePath,header=F,sep=' ')
degree_name <- paste0('degree',i,'_',j,'nodes')
colnames(degree) <- c("scanid",degree_name)
densityPath <- paste0('/data/densityLausanne',j,'.txt')
density <- read.csv(degreePath,header=F,sep=' ')
density_name <- paste0('density',i,'_',j,'nodes')
colnames(density) <- c("scanid",density_name)
resEff <- read.csv(filepath,header=F,sep=' ')
resEff[2:ncol(resEff)] <- 1/resEff[2:ncol(resEff)]
is.na(resEff) <- do.call(cbind,lapply(resEff, is.infinite))
resEff <- cbind(resEff[1],rowMeans(resEff[2:ncol(resEff)],na.rm=TRUE))
column_new <- paste0('resEff',i,'_',j,'nodes')
colnames(resEff) <- c('scanid',column_new)
resEff <- merge(resEff, degree, by=c("scanid"))
resEff <- merge(resEff, density,by=c("scanid"))
QA_df<- merge(QA_df,resEff,by=c('scanid'))
QA_minusOutlier <- QA_df[!(QA_df[[column_new]] < mean(QA_df[[column_new]]) - sd(QA_df[[column_new]])*3 | QA_df[[column_new]] > mean(QA_df[[column_new]])+ sd(QA_df[[column_new]])*3),]
placeholder <- QA_minusOutlier[[column_new]]
placeholder2 <- QA_minusOutlier[[degree_name]]
placeholder3 <- QA_minusOutlier[[density_name]]
placeholder4 <- QA_minusOutlier$ageAtScan1
placeholder5 <- QA_minusOutlier$sex
placeholder6 <- QA_minusOutlier$dti64MeanRelRMS
if (i==999)
{
df1 <- as.data.frame(cbind(rep(j,length(placeholder)),rep(0.1,length(placeholder)),placeholder,placeholder2,placeholder3, placeholder4,placeholder5,placeholder6))
} else{
df1 <- as.data.frame(cbind(rep(j,length(placeholder)),rep(100-(as.numeric(i)*10),length(placeholder)),placeholder,placeholder2,placeholder3, placeholder4,placeholder5,placeholder6))
}
if (j==83 & i==999)
{
df2 <- df1
} else{
df2 <- rbind(df2,df1)
}
}
}
df2 <- as.data.frame(df2)
colnames(df2) <- c('nodes','distortion','resources','degree','density','age','sex','motion')
df2$distortion <- as.factor(df2$distortion)
df2$sex <- as.factor(df2$sex)
plotDf <- df2
lm1 <- gam(resources~nodes+degree+density+age+sex+age*sex+motion,method="REML",data=df2)
require(scales)
ggplot(df2,aes(nodes,resources,col=distortion))+geom_point(shape=21,size=.1)+geom_jitter(width=.5)+geom_smooth(method='lm')+theme_classic(base_size=20)
ggsave('figures/fig6/highFidelityRegime.eps',device='eps',width=7.18,height=6.31)
##################################
# Glasser atlas for LOW fidelity #
##################################
# load path complexity
df1 <- read.csv('/data/pathSize.txt',sep=' ',header=F)
colnames(df1)<- c('scanid','pathComplexity')
QA_df <- merge(QA_df, df1,by='scanid')
# Res Eff Glasser
for (i in c(1,2,3,4,5,6,7,8,9,92,94,96,98,999))
{
filepath <- paste0('/data/glasser_resource_efficiency0',i,'.txt')
resEff <- read.csv(filepath,header=F,sep=' ')
resEff[2:ncol(resEff)] <- 1/resEff[2:ncol(resEff)]
is.na(resEff) <- do.call(cbind,lapply(resEff, is.infinite))
resEff <- cbind(resEff[1],rowMeans(resEff[2:ncol(resEff)],na.rm=TRUE))
colnames(resEff) <- c('scanid',"resources")
resEff <- resEff[!(resEff$resources < mean(resEff$resources) - sd(resEff$resources)*3 | resEff$resources > mean(resEff$resources)+ sd(resEff$resources)*3),]
resEff <- merge(resEff, QA_df, by='scanid')
resEff <- resEff[which(resEff[,2]>0),]
if (i==999)
{
resEff$probability <- rep(i/1000,length(resEff[,1]))
} else if (i>90 & i <100) {
resEff$probability <- rep(i/100,length(resEff[,1]))
} else {
resEff$probability <- rep(i/10,length(resEff[,1]))
}
resEff <- as.data.frame(cbind(resEff$scanid,resEff$resources,resEff$pathComplexity,resEff$probability))
colnames(resEff)<- c('scanid',"resources","pathComplexity","probability")
assign(paste0("df",i),as.data.frame(resEff))
}
df1 <- rbind(df1,df2,df3,df4,df5,df6,df7,df8,df9,df92,df94,df96,df98,df999)
df1$probability <- as.factor(df1$probability)
df2<- subset(df1,df1$probability==0.999)
df2 <- merge(df2, QA_df, by=c('scanid'))
gam1 <- gam(resources~ s(pathComplexity.x, k=4)+degree+density+s(ageAtScan1,k=4) +sex+s(ageAtScan1, by=sex, k=4) +meanMotion,fx=TRUE,method="REML",data=df2)
gam2 <- gam(resources~ pathComplexity.x+degree+density+s(ageAtScan1,k=4) +sex+s(ageAtScan1, by=sex, k=4) +meanMotion,fx=TRUE,method="REML",data=df2)
# Compare non-linear fit with linear fit: lower scores indicate better model
AIC(gam1)-AIC(gam2)
BIC(gam1)-BIC(gam2)
visreg(gam1,"pathComplexity.x",gg=T)+theme_classic(base_size=20)
ggsave('figures/fig6/lowFidelityRegime999.eps',device='eps',width=6.31,height=6.31)