-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathpreprocessingExperiments.R
178 lines (144 loc) · 6.48 KB
/
preprocessingExperiments.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
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
require(signal)
require(robfilter)
library(xts)
set.seed(0)
silencingPeriod <- 15
# define PB sandbox params
pbParams <- list()
pbParams$filterCoeff_b <- c(0.1652,0.2617,0.1652) #moving average coeff
pbParams$filterCoeff_a <- c(1.0000,-0.6671,0.5034) #autoregressive coeff
pbParams$plots <- 'sortAndFit_img' # {sortAndFit_img, sortAndFit, rgb_hsv, filter, ts} comparative plots to show
pbParams$deriv <- 1 # {0,1,2} = {x, d(x), d^2(x)}
pbParams$analyte <- 'K.pneumoniae'
pbParams$trialIdx <- 15
# define spots to be excluded
excludedSpots <- c(1, 10, 12, 19, 62, 71, 80)
# define RGB indices
excludedIndicatorsB <- excludedSpots*3
excludedIndicatorsG <- (excludedIndicatorsB -1)
excludedIndicatorsR <- (excludedIndicatorsG -1)
# assemble parameter list for manual road testing - these will be passed on automatically
parameters <- list()
parameters$centralRepoRoot <- '/Users/michaelwalton/workspace/growthDetectionExperiments'
parameters$selectedStudy <- '2015-03--Sepsis-TF'
parameters$spotLightVersion <- '03-30-2015_15-14-06'
# go to root
setwd(parameters$centralRepoRoot)
# go to study
setwd(parameters$selectedStudy)
if(parameters$selectedStudy=='2014-04--Sepsis-TF'){
parameters$inputFile <- '2014-12-23_11-27-15_2014-04--Sepsis-TF_YC.csv'
parameters$spotLightVersion <- '12-17-2014_12-27-56'
} else if (parameters$selectedStudy=='2015-03--Sepsis-TF'){
parameters$inputFile <- '2015-05-21_09-35-46_2015-03--Sepsis-TF_MW.csv'
parameters$spotLightVersion <- '03-30-2015_15-14-06'
}
metadata <- read.table(paste(parameters$inputFile, sep='/'), sep=',', header=TRUE, quote='"', comment.char="")
metadata <- metadata[which(metadata$ArrayType == "SLII-156"),]
metadata <- metadata[-which(metadata$Ignore == TRUE),]
folderStrings <- metadata$Folder
folderStrings <- gsub("\\\\","/",folderStrings)
folderStrings <- gsub("^/", "", folderStrings)
metadata$Folder <- folderStrings
rm(folderStrings)
doHsv <- function(csaData, maxColorValue = 4096) {
hsvMatrix <- matrix(0, nrow=(dim(csaData)[1]), ncol=dim(csaData)[2])
for(t in seq(dim(csaData)[1])) {
# reshape to 3 x 80, where rows of rgb are r,g,b components
rgb <- matrix(csaDataCropped[t,], nrow=3)
# map RGB ~> HSV color space
hsv <- rgb2hsv(rgb, maxColorValue = maxColorValue)
hsvMatrix[t,] <- matrix(hsv, ncol = dim(hsvMatrix)[2])
}
return(hsvMatrix)
}
doFilter <- function(csaData, a, b){
filtered <- matrix(0, nrow=(dim(csaData)[1]), ncol=dim(csaData)[2])
for(c in seq(dim(csaData)[2])){
filtered[,c] <- filter(filt=b, a=a, x=csaData[,c])
}
return(filtered)
}
doDeltaI_I0 <- function(csaData) {
refImage <- csaData[1,]
dIoIMatrix <- t(csaData)-refImage
dIoIMatrix <- t(dIoIMatrix/refImage)
return(dIoIMatrix)
}
doDeriv <- function(csaData, order = 1) {
if(order == 0) {
return(csaData)
}
csaData <- doDeriv(diff(csaData), order - 1)
}
selectedTrial <- metadata[which(metadata$Analyte == pbParams$analyte),][pbParams$trialIdx,]
raw <- read.csv(file=paste(selectedTrial$Folder, parameters$spotLightVersion, 'spots.txt', sep='/'),sep="\t",header=1)
rawMat <- as.matrix(sapply(raw, as.numeric))
# make sure additional NA columns are stripped
csaData <- rawMat[which(is.na(apply(rawMat,1, sum))==FALSE),]
csaDataCropped <- rawMat[,-1]
csaDataCropped <- csaDataCropped[,-c(excludedIndicatorsR,excludedIndicatorsG,excludedIndicatorsB)]
rgbData <- csaDataCropped
hsvData <- doHsv(csaDataCropped)
# dI/I0 >> slope >> filter
dsfRgb <- doFilter(doDeriv(doDeltaI_I0(rgbData), pbParams$deriv), pbParams$filterCoeff_a,pbParams$filterCoeff_b)
dsfHsv <- doFilter(doDeriv(doDeltaI_I0(hsvData), pbParams$deriv), pbParams$filterCoeff_a,pbParams$filterCoeff_b)
# raw
rawRgb <- doDeriv(doDeltaI_I0(rgbData), pbParams$deriv)
rawHsv <- doDeriv(doDeltaI_I0(hsvData), pbParams$deriv)
if (pbParams$plots == 'filter') {
end <- dim(dsfRgb)[1]
for(c in seq(dim(dsfRgb)[2])){
plot(raw[silencingPeriod:end,c], type='l',col='red')
lines(dsfRgb[silencingPeriod:end,c], type='l',col='blue')
legend("topright", legend=c("Raw", "Filtered"), cex=0.75, lwd=2, col=c("red", "blue"))
}
} else if (pbParams$plots == 'rgb_hsv') {
end <- dim(dsfHsv)[1]
for(i in seq(dim(dsfHsv)[2] / 3)) {
idx <- (i - 1) * 3
hsvlims <- c(min(dsfHsv[,idx + c(1,2,3)]), max(dsfHsv[,idx + c(1,2,3)]))
rgblims <- c(min(dsfRgb[,idx + c(1,2,3)]), max(dsfRgb[,idx + c(1,2,3)]))
# Multi-plot; add extra space to right of plot area; change clipping to figure
attach(mtcars)
par(mfrow=c(2,1))
# PLOT HSV
plot(dsfHsv[,idx + 1], type='l',col='darkorange', ylim = hsvlims, main = paste('Indicator #', i))
lines(dsfHsv[,idx + 2], type='l',col='purple')
lines(dsfHsv[,idx + 3], type='l',col='black')
legend("bottomright", legend=c("Hue", "Saturation", "Value"), cex=0.5, lwd=2, col=c("darkorange", "purple","black"))
# PLOT RGB
plot(dsfRgb[,idx + 1], type='l',col='red', ylim = rgblims)
lines(dsfRgb[,idx + 2], type='l',col='darkgreen')
lines(dsfRgb[,idx + 3], type='l',col='blue')
legend("bottomright", legend=c("Red", "Green", "Blue"), cex=0.5, lwd=2, col=c("red", "darkgreen","blue"))
}
} else if (pbParams$plots == 'sortAndFit' || pbParams$plots == 'sortAndFit_img') {
greyScale <- grey(seq(0, 1, length = 256))
indicatorIndex <- seq(dim(dsfRgb)[2])
regressionSlope <- matrix(nrow = dim(dsfRgb)[1], ncol = 2)
for(t in seq(dim(dsfRgb)[1])) {
rgbSorted <- dsfRgb[,order(dsfRgb[t,])]
hsvSorted <- dsfHsv[,order(dsfHsv[t,])]
rgbFit <- lm(rgbSorted[t,] ~ indicatorIndex)
hsvFit <- lm(hsvSorted[t,] ~ indicatorIndex)
regressionSlope[t,1] <- rgbFit$coefficients[1]
regressionSlope[t,2] <- hsvFit$coefficients[1]
if (pbParams$plots == 'sortAndFit_img') {
img <- as.matrix(rgbSorted[silencingPeriod:t,])
image(img, main = paste('Indicators sorted by t = ', t), col = greyScale)
}
}
plot(regressionSlope[silencingPeriod:dim(regressionSlope)[1],1], type='l', col = 'red')
#plot(regressionSlope[,2], type='l', col = 'blue')
} else if (pbParams$plots == 'ts') {
ts.plot(dsfHsv, col=rainbow(dim(dsfHsv)[2]))
ts.plot(dsfRgb, col=rainbow(dim(dsfRgb)[2]))
} else if (pbParams$plots == 'histo') {
inputXLim <- c(min(dsfRgb), max(dsfRgb))
inputYLim <- c(0,dim(dsfRgb)[2])
for(t in seq(dim(dsfRgb)[1])) {
hist(dsfRgb[t,], breaks = 10, main = "RGB Indicator Distribution",
xlab = paste(pbParams$mode, '[', t, ']'), ylab = 'n indicators', xlim = inputXLim, ylim = inputYLim)
}
}