-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathRead_LC_NSCLC_TMA_88_B.Rmd
277 lines (200 loc) · 9.24 KB
/
Read_LC_NSCLC_TMA_88_B.Rmd
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
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
---
title: "Read in CP data, NSCLC TMA 88_B"
author: "Lena Cords"
output: html_notebook
---
```{r, import libraries}
library(igraph)
library(SingleCellExperiment)
library(S4Vectors)
library(stringr)
library(DT)
```
```{r, Set wd}
#set working directory
wd <- dirname(dirname(dirname(getwd())))
wd <-"/mnt"
data_folder <- file.path(wd,"lena_processed","NSCLC_TMA","ImcSegmentationPipeline","analysis","LC_NSCLC_TMA_88_B",'cpinp')
panel_folder <- file.path(wd,"lena_processed","NSCLC_TMA","ImcSegmentationPipeline","analysis","LC_NSCLC_TMA_88_B",'cpout')
results_folder <-(file.path(wd,"lena_processed2","NSCLC_NEW","sce_objects","RAW"))
```
```{r, Data import}
#Read in: Cells, Image, Object and Panel
cells <- read.csv(file=file.path(data_folder,paste('Cells.csv')))
image <- read.csv(file=file.path(data_folder, paste("Image.csv")))
object_relationship <- read.csv(file=file.path(data_folder, paste("distance20_Object relationships.csv")))
panel <- read.csv(file=file.path(panel_folder, paste('panel.csv')))
```
```{r, What has been measured for each cell}
#number of cells
nrow(cells)
unique(sub("_c[0-9]*$", "", colnames(cells)))
#Select only the measurements you want to work on, based on the measurements displayed above Intensity_MeanIntensityCellsCorr_ColdStack
counts <- cells[, grepl("Intensity_MeanIntensityCorrected_FullStackFiltered", colnames(cells))]
```
```{r Scale counts according to CP scaling factor}
#when using 16bit data this should be 2^16 =65536
scaling_factor <- unique(image$Scaling_FullStack)
#Scale up the counts according to scaling factor from CellProfiler
counts <- counts * scaling_factor
#make sure the counts have actually been multiplied by the scaling factor
range(counts)
```
```{r Add cell metadata as S4 Object for the SCE}
#the S4Vector is needed for the SCE and a dataframe that stores the metadata per category
cell_meta <- DataFrame(CellNumber = cells$ObjectNumber,
Center_X=cells$Location_Center_X,
Center_Y=cells$Location_Center_Y,
Area = cells$AreaShape_Area,
MajorAxisLength=cells$AreaShape_MajorAxisLength,
MinorAxisLength=cells$AreaShape_MinorAxisLength,
Compartment=cells$Intensity_MeanIntensity_TumourMask_Distance_B100)
cell_meta$ImageNumber <- cells$ImageNumber
#Create DataFrame for image meta data. Then match image and cell metadata by ImageNumber
image_meta <- DataFrame(Area_Description = image$Metadata_description)
#run this to get the image number for the meta data
image_meta$ImageNumber <- image$ImageNumber
#merge cell meta and image metadata by Image Number here
#cell_meta <- merge(cell_meta, image_meta, by="ImageNumber")
cell_meta <- merge(cell_meta, image_meta, by="ImageNumber")
```
```{r, Add cell metadata as S4 Object for the SCE from file name}
#get info such as Batch, sample, ROI from original .mcd filename and add it to the cell metadata (colDat)
#this is a good overview to get the different parts of metadata from the file name
ac_info <- str_split(image$FileName_FullStack, '_', simplify=TRUE)
image$Metadata_Description
head(ac_info)
cell_meta$BatchID <- ac_info[cell_meta$ImageNumber,1]
cell_meta$Panel <- ac_info[cell_meta$ImageNumber,3]
cell_meta$TmaID <- ac_info[cell_meta$ImageNumber,5]
cell_meta$TmaBlock <- ac_info[cell_meta$ImageNumber,6]
#cell_meta$ROI <- image$Metadata_roiid[cell_meta$ImageNumber]
cell_meta$acID <- image$Metadata_acid[cell_meta$ImageNumber]
#colnames(cells)
```
```{r Set rownames for SCE}
#Should usually be cell id. However, in the cells.csv file the object number is counted per imgage, so to get unique cell IDs, they must be establishes by combining ObjectNumber, ImageNumber/ROINumber and SampleID
rownames(cell_meta) <-paste(cell_meta$TmaID,cell_meta$TmaBlock, cell_meta$acID, cell_meta$CellNumber, sep='_')
```
```{r, Match channel names with metal names in exact order}
library(DT)
#loads panel
DT::datatable(panel)
#exactl order of channels can be extracted from the _full.csv files
#header = FALSE is important here as otherwise it'll take the first metal as a header
channel_metal <- read.csv(file.path(data_folder, paste("full_channelmeta.csv")), header=FALSE)
#this reorders the panel according to the correct metal lsit
panel <- panel[match(channel_metal[,1], panel$Metal.Tag),]
#panel <- as.data.table(panel)
#use Target from panel as the correct channel name to be displayed
rownames(panel) <-panel$Clean_Target
#channels in cell file are not ordered correctly (starting with 1,11,12 ... 2,21,22 ... 3,31,32...), hence we need to reorder them
channelNumber <- as.numeric(sub("^.*_c", "", colnames(counts)))
# Order counts based on channel number
counts <- counts[,order(channelNumber, decreasing = FALSE)]
range(counts)
```
```{r build igraph object}
library(igraph)
# Construct neighbour data.frame
# First in the ImageNumber_ObjectNumber format
cur_df <- data.frame(CellID_1 = paste0(object_relationship$First.Image.Number,
"_", object_relationship$First.Object.Number),
CellID_2 = paste0(object_relationship$Second.Image.Number,
"_", object_relationship$Second.Object.Number))
# Create simple cell IDs
cellID <- paste0(cell_meta$ImageNumber, "_", cell_meta$CellNumber)
# Change cell IDs
cur_df$CellID_1 <- rownames(cell_meta)[match(cur_df$CellID_1, cellID)]
cur_df$CellID_2 <- rownames(cell_meta)[match(cur_df$CellID_2, cellID)]
# Build graph
g <- graph_from_data_frame(cur_df)
g
```
```{r Create SingleCellExperiment}
library(SingleCellExperiment)
sce_88_B <-SingleCellExperiment(assays = list(counts=t(counts)))
#marker name as row name cellID as colname
rownames(sce_88_B)<-rownames(panel)
colnames(sce_88_B)<-rownames(cell_meta)
#store metadata in colDat and everything accordingly
colData(sce_88_B) <-cell_meta
rowData(sce_88_B) <-panel
metadata(sce_88_B)<-list(graph=g)
sce_88_B$CellID <- paste(sce_88_B$TmaID,sce_88_B$TmaBlock, sce_88_B$acID, sce_88_B$CellNumber, sep='_')
#save SCE object
saveRDS(sce_88_B, file=file.path(results_folder, paste("sce_88_B_2022.rds")))
```
#add counts and clinical data
```{r, add transformed counts}
censor_val <-0.999
censor_dat <- function(x, quant = 0.999){
q = stats::quantile(x, quant)
x[x>q] = q
return(x)
}
fun.censor <- function(x) censor_dat(x, censor_val)
fun.scale <- function(x) y <- x/max(x)
#censored counts
assay(sce_88_B, "c_counts") <- t(apply(assay(sce_88_B, "counts"), 1, fun.censor))
#censored counts scaled 0-1
assay(sce_88_B, "c_counts_scaled") <- t(apply(assay(sce_88_B, "c_counts"),
1, fun.scale))
assay(sce_88_B, "c_counts_scaled")[assay(sce_88_B, "c_counts_scaled") < 0] <- 0
#censored counts asinh scaled
cofactor <- 1
assay(sce_88_B, "c_counts_asinh") <- asinh((assay(sce_88_B,"c_counts"))/cofactor)
#censored counts_asinh_scaled
assay(sce_88_B, "c_counts_asinh_scaled") <- t(apply(assay(sce_88_B, "c_counts_asinh"),
1, fun.scale))
assay(sce_88_B, "c_counts_asinh_scaled")[assay(sce_88_B, "c_counts_asinh") < 0] <- 0
results_folder <-(file.path(wd,"lena_processed2","NSCLC_NEW","sce_objects","counts"))
saveRDS(sce_88_B,file=file.path(results_folder, paste("sce_88_B_counts_RAW",".rds",sep="")))
```
#Merge ROI xy Data into sce
```{r, load clinical and meta data}
wd <-"/mnt"
data_folder <- file.path(wd,"lena_processed","NSCLC_TMA","ImcSegmentationPipeline","analysis","LC_NSCLC_TMA_88_B",'cpinp')
cd_folder <- file.path(wd,"lena_processed2","NSCLC_NEW","clinical_data")
ac_meta <- read.csv(file=file.path(data_folder, paste("acquisition_metadata.csv")))
clinical.data <-read.csv(file=file.path(cd_folder, paste("clinical_data.csv")))
clinical.data$X <-NULL
clinical.data$TmaBlock <-NULL
```
```{r, Add Roi coordinates}
ac_sub <- data.frame(ac_meta$description,ac_meta$id)
colnames(ac_sub) <- c("ROI","acID")
ac_sub$ROI[duplicated(ac_sub$ROI)]
ac_sub <- ac_sub %>%
separate(ROI,
into = c("TMA", "ROI_xy"),
sep = "(?<=[A-Za-z])(?=[0-9])",
remove=F
)
ac_sub$TMA <-NULL
ac_sub$RoiID <- paste("88",ac_sub$ROI, sep="_")
ac_sub$ROI <-NULL
head(ac_sub)
length(unique(ac_sub$ROI_xy))
length(unique(ac_sub$RoiID))
length(unique(ac_sub$acID))
ac_sub$ROI_xy[duplicated(ac_sub$ROI_xy)]
ac_sub$TMA <-"88B"
ac_sub$Tma_ac <-paste(ac_sub$TMA, ac_sub$acID, sep="_")
results_folder <-(file.path(wd,"lena_processed2","NSCLC_NEW","sce_objects","clinical_data"))
write.csv(ac_sub, file=file.path(results_folder, "ac_sub_88B.csv"))
```
```{r, add clinical data and RoiID}
#change ROI to Image ID
#names(colData(sce))[names(colData(sce)) == 'ROI'] <- 'ImageID'
head(colData(sce_88_B))
ac_clinical <- left_join(ac_sub, clinical.data, by="RoiID")
cur_DF <- as_tibble(colData(sce_88_B)) %>% left_join(ac_clinical, by = "acID") %>% DataFrame()
all.equal(paste(cur_DF$ImageNumber, cur_DF$CellNumber), paste(sce_88_B$ImageNumber, sce_88_B$CellNumber))
colData(sce_88_B) <- cur_DF
rownames(colData(sce_88_B)) <-sce_88_B$CellID
```
```{r, save sce with counts and clinical data to folder counts_clinical data}
results_folder <-(file.path(wd,"lena_processed2","NSCLC_NEW","sce_objects","counts_clinical data"))
saveRDS(sce_88_B,file=file.path(results_folder, paste("sce_88_B_counts_clinical-data_RAW",".rds",sep="")))
```