Question

I have data from an image processing program that I need to piece back together. My image data are originally an array of 18 (6 x 3) wells on a microscope slide, I need to number these wells consistently so I can identify what was in each well later on. I have the approximate x and y positions (in pixels) of the wells, and I want to sort them from left to right and then top to bottom (seems simplest) and number them 1 to 18.

like this:

     [,1] [,2] [,3] [,4] [,5] [,6]
[1,]    1    2    3    4    5    6
[2,]    7    8    9   10   11   12
[3,]   13   14   15   16   17   18

The caveats:

-Not all wells were picked up by the image processing program, so sometimes there are less than 18.

-Not all wells in the same row or column are on the exact same x or y, they seem to float around within 20-40 pixels.

-Not all images are centered on the same point or at the exact same magnification so well positions are not reliable enough to code for a range.

Example of data:

   Well BBXLeft BBYTop
1     0      39    637
2     1      43   1218
3     2     596    630
4     3     610   1212
5     4    1161    633
6     5    1164   1207
7     6    1710    623
8     7    1715   1202
9     8    2267    620
10    9    2271   1199
11   10    2824    617
12   11    2845   1197
13   12      35     57
14   13     593     53
15   14    1709     45
16   15    2262     41
17   16    2820     38

Or in reproducible form from dput:

wells <- structure(list(Well = 0:16, BBXLeft = c(39L, 43L, 596L, 610L,1161L, 1164L, 1710L, 1715L, 2267L, 2271L, 2824L, 2845L, 35L, 593L, 1709L, 2262L, 2820L), BBYTop = c(637L, 1218L, 630L, 1212L, 633L, 1207L, 623L, 1202L, 620L, 1199L, 617L, 1197L, 57L, 53L, 45L, 41L, 38L)), .Names = c("Well", "BBXLeft", "BBYTop"), class = "data.frame", row.names = c(NA, -17L))

Wells are out of order from image analysis program
This example missing well #15 (or 14 if count from 0)

Ideal output would be a new column with the left-right, top-bottom well numbers (or even “battleship” coordinates)

Sorry if this is a bit of a homework question but I really don’t know how to do this efficiently without hard-coding the cutoffs. I’m using R because it’s the only language I have a grasp on and I have other code ready to go written around this problem.

Answer 1

The code below will help you to sort your wells according to the pixel location.
However, perhaps more importantly is the following —
These are your WellNumber labels plotted against the pixel location. Is this labeling scheme intentional, or are the well locations getting shuffled when labeling the image data?

---

In order to sort your data, we simply need a natural break, such as image size. In the example below I chose 500, but you can adjust as needed.

# sort & plot the pixel corners, to get an idea for where the boundaries re
sort(wells$BBXLeft)
sort(wells$BBYTop)

plot(x=wells$BBXLeft, y=wells$BBYTop)

# add an NA for the missing value (we cant just ignore it)
wells[18,] <- c(NA, 1100, 30) 

imgSize <- 500
nrows <- 3
ncols <- 6

# Assign row and col number based on boundaries set every 'imgSize'
#  Note that rows are reversed, top to bottom
wells$row <- cut(wells$BBYTop,  breaks=seq(imgSize*nrows, 0, -imgSize), label=1:nrows)
wells$col <- cut(wells$BBXLeft, breaks=seq(0, imgSize*ncols, imgSize), label=1:ncols)

# your battleship coordinates
wells$battleship <- paste0("(", wells$row, ", ", wells$col, ")")


# then to sort it,  sorting by rows, then by cols
orderedWells <- wells$WellNo[order(wells$row, wells$col)]

# if you want to lay it out nicely, use a matrix
matrix(orderedWells, nrow=nrows, ncol=ncols, byrow=TRUE)

 #          [,1] [,2] [,3] [,4] [,5] [,6]
 #     [1,]   12   13   NA   14   15   16
 #     [2,]    0    2    4    6    8   10
 #     [3,]    1    3    5    7    9   11