• dplyr is a package for manipulating data in R
• It is an essential tool for working with ROI structural data
• Requires that you get your data into a data frame before starting

• magrittr is a package that is very useful along with dplyr
• “The magrittr package offers a set of operators which promote semantics that will improve your code”
• install.packages("magrittr")
• Here’s what it does. It converts code like this:

a = f(b)
a = g(a, c)
a = h(a, d)

with code like this

a = f(b) %>% g(c) %>% h(d)

• Back to dplyr. This is a package for manipulating and wrangling data frames
• dplyr includes a collection of verbs for manipulating data, we’ll go through them one at a time
• Here’s a link to the best docs
• The main ones for me are: filter, select, mutate, summarise

• We’ll work with some example data in the github repo under data/
• Contains two subjects’ from the Kirby21 dataset

dir("../data")

[1] "kirby127a_3_1_ax_283Labels_M2_corrected_stats.txt"
[2] "kirby142a_3_1_ax_283Labels_M2_corrected_stats.txt"

## Only if you don't have the package installed
#source("https://neuroconductor.org/neurocLite.R")
#neuro_install("MRIcloudT1volumetrics")
require(devtools)
devtools::install_github("bcaffo/MRIcloudT1volumetrics")

library(magrittr); library(dplyr); library(MRIcloudT1volumetrics) 
roiDir = "../data/"
fileList = dir(roiDir)
dat = readSubject(paste(roiDir, fileList[1], sep = "")) %>% 
    subject2df()

head(dat, 10)

                  rawid              roi volume min max     mean     std
1  kirby127a_3_1_ax.img  Telencephalon_L 531111   0 374 128.3013 51.8593
2  kirby127a_3_1_ax.img  Telencephalon_R 543404   0 300 135.0683 53.6471
3  kirby127a_3_1_ax.img   Diencephalon_L   9683  15 295 193.5488 32.2733
4  kirby127a_3_1_ax.img   Diencephalon_R   9678  10 335 193.7051 32.7869
5  kirby127a_3_1_ax.img    Mesencephalon  10268  55 307 230.8583 29.2249
6  kirby127a_3_1_ax.img    Metencephalon 159402   2 299 138.5200 52.2241
7  kirby127a_3_1_ax.img   Myelencephalon   4973  12 286 199.8497 36.6501
8  kirby127a_3_1_ax.img              CSF 109776   0 258  33.0193 26.3262
9  kirby127a_3_1_ax.img CerebralCortex_L 276965   0 249  88.5014 27.7510
10 kirby127a_3_1_ax.img CerebralCortex_R 282858   0 295  93.7283 28.1726
   type level
1     1     1
2     1     1
3     1     1
4     1     1
5     1     1
6     1     1
7     1     1
8     1     1
9     1     2
10    1     2

• Minus removes columns
• Otherwise select will keep the columns that you specify

dat = select(dat, -rawid)
head(dat, 10)

                roi volume min max     mean     std type level
1   Telencephalon_L 531111   0 374 128.3013 51.8593    1     1
2   Telencephalon_R 543404   0 300 135.0683 53.6471    1     1
3    Diencephalon_L   9683  15 295 193.5488 32.2733    1     1
4    Diencephalon_R   9678  10 335 193.7051 32.7869    1     1
5     Mesencephalon  10268  55 307 230.8583 29.2249    1     1
6     Metencephalon 159402   2 299 138.5200 52.2241    1     1
7    Myelencephalon   4973  12 286 199.8497 36.6501    1     1
8               CSF 109776   0 258  33.0193 26.3262    1     1
9  CerebralCortex_L 276965   0 249  88.5014 27.7510    1     2
10 CerebralCortex_R 282858   0 295  93.7283 28.1726    1     2

• filter selects rows of a dataframe based on criteria

dat  = filter(dat, type == 1, level == 1)
dat

              roi volume min max     mean     std type level
1 Telencephalon_L 531111   0 374 128.3013 51.8593    1     1
2 Telencephalon_R 543404   0 300 135.0683 53.6471    1     1
3  Diencephalon_L   9683  15 295 193.5488 32.2733    1     1
4  Diencephalon_R   9678  10 335 193.7051 32.7869    1     1
5   Mesencephalon  10268  55 307 230.8583 29.2249    1     1
6   Metencephalon 159402   2 299 138.5200 52.2241    1     1
7  Myelencephalon   4973  12 286 199.8497 36.6501    1     1
8             CSF 109776   0 258  33.0193 26.3262    1     1

• Let’s create an ICV column

dat = mutate(dat, icv = sum(volume))
dat

              roi volume min max     mean     std type level     icv
1 Telencephalon_L 531111   0 374 128.3013 51.8593    1     1 1378295
2 Telencephalon_R 543404   0 300 135.0683 53.6471    1     1 1378295
3  Diencephalon_L   9683  15 295 193.5488 32.2733    1     1 1378295
4  Diencephalon_R   9678  10 335 193.7051 32.7869    1     1 1378295
5   Mesencephalon  10268  55 307 230.8583 29.2249    1     1 1378295
6   Metencephalon 159402   2 299 138.5200 52.2241    1     1 1378295
7  Myelencephalon   4973  12 286 199.8497 36.6501    1     1 1378295
8             CSF 109776   0 258  33.0193 26.3262    1     1 1378295

Everything we’ve done so far

dat = readSubject(paste(roiDir, fileList[1], sep = "")) %>% 
    subject2df() %>%
    filter(type == 1, level == 1) %>%
    mutate(icv = sum(volumes))

fullData = readSubject(paste(roiDir, fileList[1], sep = "")) %>% 
        subject2df()
head(fullData)

                 rawid             roi volume min max     mean     std
1 kirby127a_3_1_ax.img Telencephalon_L 531111   0 374 128.3013 51.8593
2 kirby127a_3_1_ax.img Telencephalon_R 543404   0 300 135.0683 53.6471
3 kirby127a_3_1_ax.img  Diencephalon_L   9683  15 295 193.5488 32.2733
4 kirby127a_3_1_ax.img  Diencephalon_R   9678  10 335 193.7051 32.7869
5 kirby127a_3_1_ax.img   Mesencephalon  10268  55 307 230.8583 29.2249
6 kirby127a_3_1_ax.img   Metencephalon 159402   2 299 138.5200 52.2241
  type level
1    1     1
2    1     1
3    1     1
4    1     1
5    1     1
6    1     1

• Question, what’s the volumetric sum at each level and type?
• summarise creates summaries
• group_by operates functions over groups

fullData %>% group_by(type, level) %>% 
    summarise(nvol = n(), sum = sum(volume), mean = mean(volume)) %>%
    as.data.frame() #just for display

   type level nvol     sum       mean
1     1     1    8 1378295 172286.875
2     1     2   19 1378299  72542.053
3     1     3   54 1378314  25524.333
4     1     4  136 1378345  10134.890
5     1     5  276 1378405   4994.221
6     2     1    5 1378295 275659.000
7     2     2   18 1378299  76572.167
8     2     3   52 1378312  26506.000
9     2     4   70 1378318  19690.257
10    2     5  198 1378373   6961.480