Detecting Stick Slips of Cell Membranes

Data Ingestion

setwd("/Users/joergheintz/Documents/06_Projects/01_SIAM_Jump")
df<-read.xlsx("df_jump.xlsx", 2)[,c(2, 3)]
mydata<-df[513:1024,]
x<-mydata$x
y<-mydata$v.def
colnames(mydata)<-c("x", "y")

Detect Slope, Curve Turning Points and Add Knots

for (i in 1:length(mydata$y)){
        a <- mydata[i,]$y
        b <- mydata[i+1,]$y
        if (!is.na(b)) {
        if (a>b) mydata$mSL[i] = 0
        if (a<b) mydata$mSL[i] = 1
        
        }
}
#for (i in 1:length(mydata$y)){
k = length(mydata$y)
i=1
b=1
while ((i < k) & !is.na(b)) {
        a <- mydata[i,]$mSL
        b <- mydata[i+1,]$mSL
        if ( a==1 & !b == a) {
                mydata[i+1, ]$mSL = 1
                helper<-mydata[i+1, ]
                helper$mSL<-0
                mydata <- rbind(mydata, helper)
                mydata<-mydata[order(mydata$x),]
                k = length(mydata$y)
                i=i+1
        }
        if ( a==0 & !b == a) {
                mydata[i+1, ]$mSL = 0
                helper<-mydata[i+1, ]
                helper$mSL<-1
                mydata <- rbind(mydata, helper)
                mydata<-mydata[order(mydata$x),]
                k = length(mydata$y)
                i=i+1
        
        }
        i=i+1
}

mydata<-mydata[order(mydata$x),]

Determine pos/neg Slope / Group Points to Slopes

# marking positive and negative slope
mydata$gradient <- 1
mydata[mydata$mSL < 1,"gradient"]<- -1

# cluster
n=0
flipflop = 1
for (i in 1:length(mydata$y)){
        if (mydata$gradient[i] == 1)  {
              if (flipflop == 1) {n = n + 1}
                mydata$cluster[i]<-n
                flipflop=0
        }
        if (mydata$gradient[i] == -1)  {
              if (flipflop == 0) {n = n + 1}
                mydata$cluster[i]<-n
                flipflop=1
        }
}

#Calculating the slope 


xmin<-as.data.frame(tapply(mydata$x, mydata$cluster, min))
xmax<-as.data.frame(tapply(mydata$x, mydata$cluster,max))
ymin<-as.data.frame(tapply(mydata$y, mydata$cluster, min))
ymax<-as.data.frame(tapply(mydata$y, mydata$cluster, max))
sl_fac<-as.data.frame(tapply(mydata$gradient, mydata$cluster, sum))

myLines<-data.frame()
myLines<-cbind(xmin, xmax, ymin, ymax, sl_fac)
colnames(myLines)<-c("xmin", "xmax","ymin","ymax", "sl_fac")
myLines[myLines$sl_fac < 0, c("ymin", "ymax")] <- myLines[myLines$sl_fac < 0, c("ymax", "ymin")]
myLines$dx<-myLines$xmax  - myLines$xmin
myLines$dy<-myLines$ymax  - myLines$ymin
myLines$slope<-myLines$dy/myLines$dx*myLines$sl_fac/abs(myLines$sl_fac)

## peaks
peaks<-myLines[order(myLines$dy),][1:10,]

Determine max Segment Changes

# Determining the largest  segment jump 
mydata$ID<-1:length(mydata$y)
for (i in 1:length(mydata$y)){
        a <- mydata[i,]$y
        b <- mydata[i+1,]$y
        if (!is.na(b)) {
        # if (a>b) mydata$mSL[i] = 0
        # if (a<b) mydata$mSL[i] = 1
        mydata$yi1_minus_yi[i+1] = b-a 
        
        }
}

# Top 10 segment jumps
segjumps1<-mydata[order(mydata$yi1_minus_yi), ][1:10,]
segjumps2<-mydata[segjumps1$ID-1 %in% mydata$ID ,]
segjumps2<-segjumps2[ ,c("x", "y", "ID")]
colnames(segjumps1) <-c("xmax", "ymax","mSL", "gradient", "cluster","yi1_minus_yi","ID")
colnames(segjumps2) <-c("xmin", "ymin","ID")
segments<-cbind(segjumps1, segjumps2)

# Linear Regression
slope2of2<-lm(formula = df$v.def[513:1024] ~ df$x[513:1024], data = df)$coeff

Plot

1. Curve
2. Segments
3. Peaks
4. Regression Line

p <- ggplot(mydata, aes(x, y))
p = p + geom_segment(aes(x = xmin, y = ymin, xend = xmax, yend = ymax, colour = "line"), colour="lightblue", data = myLines)
p = p + geom_point(colour="blue", size = 0.02)

p

p = p + geom_segment(aes(x = xmin, y = ymin, xend = xmax, yend = ymax, colour = "peaks"), colour="green", data = peaks)
p = p + geom_segment(aes(x = xmin, y = ymin, xend = xmax, yend = ymax, colour = "segments"), colour="red", data = segments)
p

p = p + geom_abline(intercept = slope2of2[1], slope =slope2of2[2], colour = "salmon", linetype = 2)
#p = p + geom_line(colour="black", size = 0.3)
p