## Error in eval(expr, envir, enclos): object 'opts_chunk' not found
scriptName <- "SetArea_Measure_UserConfirm.ijm"
pdfName <- "Usage_of_ImageJ_macro_SetArea_Measure_UserConfirm.pdf"
resultsUser1 <- "Results_SetArea_Measure_UserConfirm_User1.txt"
resultsUser2 <- "Results_SetArea_Measure_UserConfirm_User2.txt"Images of Fucci cells loaded in the C1 capture array, were taken before cell lysis. Three modes have been used for imaging: bright field (BF), Green and Red. Raw images in the Cellomics specific C01 format were used for measuring fluorescence intensities and reporting errors in Fiji ImageJ (plain ImageJ users need to install the Bio-Formats plugin). All platforms version by using the macro SetArea_Measure_UserConfirm.ijm. The usage of this macro is illustrated in the accompanying PDF file Usage_of_ImageJ_macro_SetArea_Measure_UserConfirm.pdf. The macro will ask the user to select the directory which contains all the raw image files. The following procedures are done:
- Open images one by one: first bright field, then Green, then Red channel images
- Draw a circular area with a diameter of 13 pixels on the bright field image. For other projects, if the circle is too small, edit the macro to increase the variable
defaultDiameter, for example to 30. - Open dialogue box and and wait for the user to place the circle over the cell
- Measure fluorescence intensities on this defined area for the Green and Red channel image, and measure background intensity levels 100 x 50 pixels adjacent to the cell (unless position is changed by the user)
- Open dialogue box and ask the user for an error report: comment and error type
- Record the file name, coordinates, size of area, mean intnsity, standard deviation, minimum and maximum intensities and an Error report
- Repeat the previous steps for every set of three images in the folder
- Tip: if you change the shape of the circular area by mistake, you can
re-create a new one of diameter 13 by opening the System Clipboard
(File → New → …) and running the command
makeOval(250, 250, 13, 13)from there.
The Fiji ImageJ macro was run independently by Elo Madissoon, User 1 (file Results_SetArea_Measure_UserConfirm_User1.txt) and Michael Böttcher, User 2 (file Results_SetArea_Measure_UserConfirm_User2.txt). The current document explains the column names in the results file, analyzes the consistency of the measurements, compares and finds consensus in the error reporting and gives an overview of the final values.
Identifier of the well, which is the common part of the filename for all channels (bright field, red and green)
Pixel coordinates of the upper left corner of the smallest rectangle that completely contain the selection area.
Size of the selection area.
mean, standard deviation, minimum and maximum fluorescence intensity in the selection. Corresponding measurements to the background area (adjacent to the cell) when prefix is ".bg" and to green channel or red channel when the suffix is ".ch2" or ".ch3" correspondingly.
The error types:
- "0-No Error" (default)
- "1-No cell"
- "2-Debris"
- "3-OutOfFocus"
- "4-MultipleCells"
Write additional notes here, if necessary. Deafults to "No comment"
A boolean value is given to each cell depending on the error type (e.g. "0-No Error" corresponds to FALSE)
Load the fluorescence results files and summarise the content.
panel.cor <- function(x, y, digits=2, prefix="", cex.cor) {
usr <- par("usr"); on.exit(par(usr))
par(usr = c(0, 1, 0, 1))
r <- abs(cor(x, y))
txt <- format(c(r, 0.123456789), digits=digits)[1]
txt <- paste(prefix, txt, sep="")
if(missing(cex.cor)) cex <- 0.8/strwidth(txt)
test <- cor.test(x,y)
signif <- symnum(test$p.value, corr = FALSE, na = FALSE,
cutpoints = c(0, 0.001, 0.01, 0.05, 0.1, 1),
symbols = c("***", "**", "*", ".", " "))
text(0.5, 0.5, txt, cex = cex * r)
text(.8, .8, signif, cex=cex, col=2)
}
user1 <- read.table(resultsUser1, header=TRUE, sep="\t", row.names=1)
user2 <- read.table(resultsUser2, header=TRUE, sep="\t", row.names=1)
colnames(user2)## [1] "cell_id" "cell_coord_x" "cell_coord_y" "area" "mean_ch2" "std_ch2"
## [7] "min_ch2" "max_ch2" "bg_coord_ch2_x" "bg_coord_ch2_y" "bg_area_ch2" "bg_mean_ch2"
## [13] "bg_std_ch2" "bg_min_ch2" "bg_max_ch2" "area_ch3" "mean_ch3" "std_ch3"
## [19] "min_ch3" "max_ch3" "bg_coord_ch3_x" "bg_coord_ch3_y" "bg_area_ch3" "bg_mean_ch3"
## [25] "bg_std_ch3" "bg_min_ch3" "bg_max_ch3" "Error" "Comment"
The cell location has to be manually set for all bright field images. The correalation of the cell coordinates reflects reproducibility of the cell location determination. The correlation of fluorescence intensities reflects the reproducibility of the results. The correlation between cell coordinates and fluorescence values is plotted, color coded by Run.
subb <- substr(user1$cell_id, 1, 12)
mfrow.orig <- par(mfrow=c(1,2))
plot(user1$cell_coord_x, user2$cell_coord_x, col=as.factor(subb), main="Correlation of the x-coordinates", xlab="User 1", ylab="User 2")
legend("topleft", legend=c(levels(as.factor(subb))), pch=1, col=c(1:5), text.col=c(1:5), bty="n")
plot(user1$cell_coord_y, user2$cell_coord_y, col=as.factor(subb), main="Correlation of the y-coordinates", xlab="User 1", ylab="User 2")
legend("topleft", legend=c(levels(as.factor(subb))), pch=1, col=c(1:5), text.col=c(1:5), bty="n")
plot(user1$mean_ch2, user2$mean_ch2, main="Correlation of measures in Ch2, green", xlab="User 1", ylab="User 2")
legend("topleft", legend=c(levels(as.factor(subb))), pch=1, col=c(1:5), text.col=c(1:5), bty="n")
plot(user1$mean_ch3, user2$mean_ch3, main="Correlation of measures in Ch3, red", xlab="User 1", ylab="User 2")
legend("topleft", legend=c(levels(as.factor(subb))), pch=1, col=c(1:5), text.col=c(1:5), bty="n")
The cell loacation and fluorescence mean values correlate with each other between the two repeated measurements.
The mean, minimum, maximum and standard deviation are expected to correlate among themselves, in both green and red channel and between the two independent scorings.
pairs(cbind(user1.mean = user1$mean_ch2, user2.mean = user2$mean_ch2, user1.min = user1$min_ch2, user2.min = user2$min_ch2, user1.max = user1$max_ch2, user2.max = user2$max_ch2, user1.std = user1$std_ch2, user2.std = user2$std_ch2), lower.panel=panel.smooth, upper.panel=panel.cor, main="Comparison of values in Channel 2, green")
pairs(cbind(user1.mean = user1$mean.ch3, user2.mean = user2$mean_ch3, user1.min = user1$min.ch3, user2.min = user2$min_ch3, user1.max = user1$max_ch3, user2.max = user2$max_ch3, user1.std = user1$std_ch3, user2.std = user2$std_ch3), lower.panel=panel.smooth, upper.panel=panel.cor, main="Comparison of values in Channel 3, red")
Mean, maximum and standard deviation correlate well with each other in both channels for both scorings. The minimum values are not correlating as good. This can be explained the by lower range of values and different distribution of fluorescence values inside the cell. The middle and nucleus of the cell is more fluorescent, while the membrane of the cell does not have any fluorescence in well focused images.
Compare the intensity levels of fluroescence in respect to the error type and background.
par(mfrow <- c(2,2))## NULL
mar.orig <- par(mar=c(5.1, 10.0, 4.1, 2.1))
flBoxplot <- function (DATA){
title.2 <- paste("Fluorescence by ", deparse(substitute(DATA)),", green",sep="")
title.3 <- paste("Fluorescence by ", deparse(substitute(DATA)),", red",sep="")
y <- c(as.character(DATA[,"Error"]), rep("background", nrow(DATA)))
x.ch2 <- c(DATA[,"mean_ch2"], DATA[,"bg_mean_ch2"])
x.ch3 <- c(DATA[,"mean_ch3"], DATA[,"bg_mean_ch3"])
boxplot( x.ch2~y, main=title.2, las=1, horizontal=TRUE)
boxplot( x.ch3~y, main=title.3, las=1, horizontal=TRUE)
}
flBoxplot(user1)
flBoxplot(user2)
The low fluorescence value in wells with error 1-No cell is expected and is similar to the background levels of fluorescence in both green and red channel. Cells classified as 2-Debris could be defragmented cells, which have not lost their fluroescence, therefore the values are varying. The boxplots indicate that error report are different for 2-Debris definition between two scorings.
Observe the number of similar and different error scores between User 1 and User 2 and the number of various classifications for both scorings.
summary(user1$Error == user2$Error)## Mode FALSE TRUE NA's
## logical 44 334 0
cbind(user1.error = summary(user1$Error), user2.error = summary(user2$Error))## user1.error user2.error
## 0-No Error 277 297
## 1-No cell 52 43
## 2-Debris 7 13
## 3-OutOfFocus 22 10
## 4-MultipleCells 20 15
The error reportings differ between the two observers. Display the wells where reports differ, but at least one report has "0-No Error". Add Comment column from User 2, since there were no comments from User 1 for these cells.
# Error report differences
diff <- user1$Error!=user2$Error
differ <- cbind(cell=as.character(user1$cell_id), user1.comment = as.character(user1$Comment), user1.error = as.character(user1$Error), user2.error = as.character(user2$Error), user2.comment = as.character(substr(user2$Comment,1,45)))[diff,]
difference <- differ[(differ[,"user1.error"] == "0-No Error") | (differ[,"user2.error"] == "0-No Error"),]
levels(as.factor(difference[,"user1.comment"]))## [1] "No comment"
length(levels(as.factor(difference[,"user2.comment"])))## [1] 13
d <- data.frame(difference[order(difference[,"user1.error"]),-2])
d## cell user1.error user2.error user2.comment
## 1 1772-062-248_B04 0-No Error 2-Debris No comment
## 2 1772-062-249_C08 0-No Error 2-Debris cell looks broken
## 3 1772-062-249_C11 0-No Error 4-MultipleCells No comment
## 4 1772-062-249_E08 0-No Error 2-Debris cell does not looks damaged
## 5 1772-064-103_E03 0-No Error 2-Debris No comment
## 6 1772-067-038_G03 0-No Error 2-Debris cell not in capture area and has strange shap
## 7 1772-067-039_H12 0-No Error 4-MultipleCells No comment
## 8 1772-067-038_F09 1-No cell 0-No Error No comment
## 9 1772-067-038_H10 1-No cell 0-No Error No comment
## 10 1772-062-248_D09 2-Debris 0-No Error No comment
## 11 1772-062-249_E06 2-Debris 0-No Error No comment
## 12 1772-064-103_C01 2-Debris 0-No Error something was in background of cell; maybe a
## 13 1772-067-038_D12 2-Debris 0-No Error No comment
## 14 1772-062-248_A04 3-OutOfFocus 0-No Error green fl. slightly larger than circle
## 15 1772-062-248_C03 3-OutOfFocus 0-No Error green and red fl. larger than circle
## 16 1772-062-248_C04 3-OutOfFocus 0-No Error green and red fl. slightly larger than circle
## 17 1772-062-248_C05 3-OutOfFocus 0-No Error green and red fl. larger than circle
## 18 1772-062-248_C07 3-OutOfFocus 0-No Error green and red fl. slightly larger than circle
## 19 1772-062-248_C10 3-OutOfFocus 0-No Error no fl., but normal cell
## 20 1772-062-248_D01 3-OutOfFocus 0-No Error red fl. much larger than circle
## 21 1772-062-248_D06 3-OutOfFocus 0-No Error red fl. much larger than circle
## 22 1772-062-248_D08 3-OutOfFocus 0-No Error green and red fl. much larger than circle
## 23 1772-062-249_H03 3-OutOfFocus 0-No Error not sharp, but not too much out of focus eith
## 24 1772-067-039_D05 3-OutOfFocus 0-No Error No comment
## 25 1772-062-249_D03 4-MultipleCells 0-No Error No comment
## 26 1772-062-249_D11 4-MultipleCells 0-No Error No comment
## 27 1772-062-249_E01 4-MultipleCells 0-No Error No comment
## 28 1772-067-038_B11 4-MultipleCells 0-No Error might be more than 1 cell
## 29 1772-067-038_D04 4-MultipleCells 0-No Error No comment
## 30 1772-067-038_D06 4-MultipleCells 0-No Error No comment
## 31 1772-067-038_H01 4-MultipleCells 0-No Error No comment
## 32 1772-067-038_H02 4-MultipleCells 0-No Error No comment
## 33 1772-067-038_H07 4-MultipleCells 0-No Error No comment
## 34 1772-067-039_E07 4-MultipleCells 0-No Error No comment
There is inconsistency in the error reporting for 4 cells. A third opinion from coworkers was used to find consensus for the cells with at least one 0-No Error by User 1 or User 2. All the cells with error 4-MultipleCells by either User 1 or User 2 were decided to score as 0-No Error. Six cells with 3-OutOfFocus were decided to keep as 3-OutOfFocus and five cells re-scored as 0-No Error. 13 more cells with various errors were dicussed. The new scoring for cells with non-consensus was done as follows:
d$error <- d$user2.error
d$error <- as.matrix(d$error)
d$error[d$user2.error == "4-MultipleCells",] <- "0-No Error"
Consensus_OutOfFocus <- c("1772-062-248_C03","1772-062-248_C04","1772-062-248_C05","1772-062-248_C10","1772-062-248_D06","1772-062-248_D08")
d$error[match(Consensus_OutOfFocus, d$cell),] <- "3-OutOfFocus"
Consensus_Discard <- c("1772-062-248_B04","1772-062-249_C08","1772-062-038_G03","1772-062-249_E06","1772-062-103_C01","1772-062-038_D12")
d$error[match(Consensus_Discard, d$cell),] <- "2-Debris"
Consensus_Keep <- c("1772-062-249_C11","1772-062-249_E08","1772-062-103_E03", "1772-062-039_H12","1772-062-038_F09","1772-067-038_H10","1772-062-248_D09")
d$error[match(Consensus_Keep, d$cell),] <- "0-No Error"
d[,c(-4)]## cell user1.error user2.error error
## 1 1772-062-248_B04 0-No Error 2-Debris 2-Debris
## 2 1772-062-249_C08 0-No Error 2-Debris 2-Debris
## 3 1772-062-249_C11 0-No Error 4-MultipleCells 0-No Error
## 4 1772-062-249_E08 0-No Error 2-Debris 0-No Error
## 5 1772-064-103_E03 0-No Error 2-Debris 2-Debris
## 6 1772-067-038_G03 0-No Error 2-Debris 2-Debris
## 7 1772-067-039_H12 0-No Error 4-MultipleCells 0-No Error
## 8 1772-067-038_F09 1-No cell 0-No Error 0-No Error
## 9 1772-067-038_H10 1-No cell 0-No Error 0-No Error
## 10 1772-062-248_D09 2-Debris 0-No Error 0-No Error
## 11 1772-062-249_E06 2-Debris 0-No Error 2-Debris
## 12 1772-064-103_C01 2-Debris 0-No Error 0-No Error
## 13 1772-067-038_D12 2-Debris 0-No Error 0-No Error
## 14 1772-062-248_A04 3-OutOfFocus 0-No Error 0-No Error
## 15 1772-062-248_C03 3-OutOfFocus 0-No Error 3-OutOfFocus
## 16 1772-062-248_C04 3-OutOfFocus 0-No Error 3-OutOfFocus
## 17 1772-062-248_C05 3-OutOfFocus 0-No Error 3-OutOfFocus
## 18 1772-062-248_C07 3-OutOfFocus 0-No Error 0-No Error
## 19 1772-062-248_C10 3-OutOfFocus 0-No Error 3-OutOfFocus
## 20 1772-062-248_D01 3-OutOfFocus 0-No Error 0-No Error
## 21 1772-062-248_D06 3-OutOfFocus 0-No Error 3-OutOfFocus
## 22 1772-062-248_D08 3-OutOfFocus 0-No Error 3-OutOfFocus
## 23 1772-062-249_H03 3-OutOfFocus 0-No Error 0-No Error
## 24 1772-067-039_D05 3-OutOfFocus 0-No Error 0-No Error
## 25 1772-062-249_D03 4-MultipleCells 0-No Error 0-No Error
## 26 1772-062-249_D11 4-MultipleCells 0-No Error 0-No Error
## 27 1772-062-249_E01 4-MultipleCells 0-No Error 0-No Error
## 28 1772-067-038_B11 4-MultipleCells 0-No Error 0-No Error
## 29 1772-067-038_D04 4-MultipleCells 0-No Error 0-No Error
## 30 1772-067-038_D06 4-MultipleCells 0-No Error 0-No Error
## 31 1772-067-038_H01 4-MultipleCells 0-No Error 0-No Error
## 32 1772-067-038_H02 4-MultipleCells 0-No Error 0-No Error
## 33 1772-067-038_H07 4-MultipleCells 0-No Error 0-No Error
## 34 1772-067-039_E07 4-MultipleCells 0-No Error 0-No Error
Report from User 1 used for final fluorescence values. Prepare additional columns for Run and Well. Rename errors in Error column to match the consensus error report. Add Fluo_QC column for wells with consensus errors and outliers (fluorescence > 200).
user1$Well <- substr(user1$cell_id, 14, 16)
user1$Run <- as.factor(paste(substr(subb, 1, 4), substr(subb, 6, 8),substr(subb, 10, 13), sep="-"))
user1$Well <- as.factor(user1$Well)
user1$Error[match(d$cell, user1$cell_id)] <- d$error
user1$fluo_QC <- FALSE
user1[user1$Error != "0-No Error", "fluo_QC"] <- TRUE
user1[user1$mean_ch3 > 200, "fluo_QC"] <- TRUE
user1[user1$mean_ch2 > 200, "fluo_QC"] <- TRUE- The ImageJ macro produces reproducible results in the x and y coordinates and in the mean values of fluorescence
- The mean, standard deviation and maxium values correlate well with each other in both the green and the red channels
- A consensus was achieved with respect to the error type
- Fluorescence table was saved as
Results_fluorescence.txt - There are image files for 378 cells, and for 102 cells of the runs 1772-062-248, 1772-062-249 and 1772-064-103, some of the data was lost because of a case-sensitivity bug
summary(user1)## cell_id cell_coord_x cell_coord_y area mean_ch2 std_ch2
## 1772-062-248_A01: 1 Min. :170.0 Min. :151.0 Min. :214.6 Min. : 14.93 Min. : 3.810
## 1772-062-248_A02: 1 1st Qu.:217.2 1st Qu.:166.2 1st Qu.:214.6 1st Qu.: 23.32 1st Qu.: 5.301
## 1772-062-248_A03: 1 Median :233.5 Median :176.0 Median :214.6 Median : 29.80 Median : 8.095
## 1772-062-248_A04: 1 Mean :247.4 Mean :188.4 Mean :221.2 Mean : 43.12 Mean : 19.783
## 1772-062-248_A05: 1 3rd Qu.:259.0 3rd Qu.:206.0 3rd Qu.:228.0 3rd Qu.: 60.15 3rd Qu.: 27.661
## 1772-062-248_A06: 1 Max. :363.0 Max. :254.0 Max. :228.0 Max. :218.37 Max. :128.649
## (Other) :372
## min_ch2 max_ch2 bg_coord_ch2_x bg_coord_ch2_y bg_area_ch2 bg_mean_ch2 bg_std_ch2
## Min. : 0.00 Min. : 25.00 Min. :120.0 Min. :161.0 Min. :214.6 Min. :14.94 Min. :3.609
## 1st Qu.: 9.00 1st Qu.: 38.00 1st Qu.:168.0 1st Qu.:216.0 1st Qu.:214.6 1st Qu.:16.93 1st Qu.:4.467
## Median :12.00 Median : 52.00 Median :183.5 Median :225.5 Median :214.6 Median :21.96 Median :4.670
## Mean :12.95 Mean : 91.26 Mean :197.4 Mean :237.9 Mean :221.2 Mean :20.86 Mean :4.681
## 3rd Qu.:16.00 3rd Qu.:128.50 3rd Qu.:209.0 3rd Qu.:255.8 3rd Qu.:228.0 3rd Qu.:22.87 3rd Qu.:4.889
## Max. :67.00 Max. :488.00 Max. :313.0 Max. :304.0 Max. :228.0 Max. :27.25 Max. :5.501
##
## bg_min_ch2 bg_max_ch2 area_ch3 mean_ch3 std_ch3 min_ch3 max_ch3
## Min. : 0.00 Min. :24.00 Min. :214.6 Min. : 10.39 Min. : 3.827 Min. : 0.00 Min. : 19.0
## 1st Qu.: 6.00 1st Qu.:30.00 1st Qu.:214.6 1st Qu.: 16.70 1st Qu.: 4.872 1st Qu.: 4.00 1st Qu.: 30.0
## Median : 9.00 Median :34.00 Median :214.6 Median : 39.83 Median : 17.341 Median : 6.00 Median : 90.5
## Mean : 8.68 Mean :33.16 Mean :221.2 Mean : 47.88 Mean : 27.816 Mean : 9.68 Mean :115.1
## 3rd Qu.:11.00 3rd Qu.:36.00 3rd Qu.:228.0 3rd Qu.: 68.30 3rd Qu.: 43.501 3rd Qu.: 12.00 3rd Qu.:169.0
## Max. :17.00 Max. :42.00 Max. :228.0 Max. :238.86 Max. :147.678 Max. :116.00 Max. :515.0
##
## bg_coord_ch3_x bg_coord_ch3_y bg_area_ch3 bg_mean_ch3 bg_std_ch3 bg_min_ch3 bg_max_ch3
## Min. :120.0 Min. :157.0 Min. :214.6 Min. : 9.701 Min. :3.710 Min. : 0.000 Min. :18.00
## 1st Qu.:168.0 1st Qu.:216.0 1st Qu.:214.6 1st Qu.:12.560 1st Qu.:4.166 1st Qu.: 1.000 1st Qu.:23.00
## Median :183.5 Median :225.5 Median :214.6 Median :13.212 Median :4.351 Median : 3.000 Median :25.00
## Mean :197.4 Mean :237.9 Mean :221.2 Mean :13.797 Mean :4.351 Mean : 2.794 Mean :25.31
## 3rd Qu.:209.0 3rd Qu.:255.8 3rd Qu.:228.0 3rd Qu.:14.575 3rd Qu.:4.524 3rd Qu.: 4.000 3rd Qu.:27.00
## Max. :313.0 Max. :304.0 Max. :228.0 Max. :22.664 Max. :5.069 Max. :10.000 Max. :34.00
##
## Error Comment Well Run fluo_QC
## 0-No Error :293 BG XY change : 1 C07 : 5 1772-062-248:62 Mode :logical
## 1-No cell : 50 BG change XY : 1 C08 : 5 1772-062-249:62 FALSE:290
## 2-Debris : 8 No comment :373 C11 : 5 1772-064-103:62 TRUE :88
## 3-OutOfFocus : 17 Small cell or debris? : 2 C12 : 5 1772-067-038:96 NA's :0
## 4-MultipleCells: 10 Very small cell or debris?: 1 D01 : 5 1772-067-039:96
## D02 : 5
## (Other):348
write.table(user1, "Results_fluorescence.csv", sep=",")
sessionInfo()## R version 3.1.0 (2014-04-10)
## Platform: x86_64-apple-darwin13.1.0 (64-bit)
##
## locale:
## [1] C
##
## attached base packages:
## [1] stats graphics grDevices utils datasets base
##
## loaded via a namespace (and not attached):
## [1] evaluate_0.5.5 formatR_1.0 knitr_1.8 stringr_0.6.2 tools_3.1.0