-
Notifications
You must be signed in to change notification settings - Fork 4
/
ChIP_analysis.diff_bind_PDTX.R
322 lines (275 loc) · 13.8 KB
/
ChIP_analysis.diff_bind_PDTX.R
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
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
#' @title differential binding analysis
#' @description perform iterative differential binding analysis
#' @param processed drosophila and input-subtracted counts
#' @return differential binding sites for each PDTX
#' @author angela simeone \email{[email protected]}
# CAREFUL: output_dir_name and base_dir DEFINED IN PREVIOUS SCRIPT ChIP_analysis.normalize_data.R
base_dir <- '/PDTX_dir'
setwd(base_dir)
output_dir_name <- 'PDTX_dm6_inputSubtraction_/'
adir <- paste(base_dir,output_dir_name,sep='')
setwd(adir)
log_file_name <- paste(gsub('\\/$','',adir),'.log',sep='')
library(gplots)
library(edgeR)
library(dendextend)
library(RColorBrewer)
# ====================
# == INPUT FILES =====
# ====================
# tab-separated file file containing peak file and read counts in each peak, for each samples: see peak_example.bedgraph
processed_peak_file_counts_name <- paste(adir,'peak_counts.norm_filtered.tab',sep='')
# tab-separated file containing sample information see sample_info_example.txt
sample_file_name <- "/PDTX_dir/sample_file_PDTX_only.txt"
# tab-separated file containing the normalization factor for each sample: norm_example.txt
output_prefix <- paste(tools::file_path_sans_ext(processed_peak_file_counts_name),'.detable',sep='')
# ==============================
# ===== CONTROL PARAMETERS =====
# ==============================
save_plots <- 0
save_tables <- 1
th <- 0.01
FC <- 0.6
de_type <- 0 # 0: split; 1: merged; 2:batch
#de_type <- 2# 0: split; 1: merged; 2:batch
# ====== Load input files ======
peaks <- read.table(processed_peak_file_counts_name,stringsAsFactors = F,header=T,sep='\t')
colnames(peaks) <- gsub('^X','',colnames(peaks))
peaks <- peaks[-grep('chr.*._.*',peaks[,1]),] # to exclude those in chr_
sample_file <- read.table(sample_file_name,stringsAsFactors = F,header=F,sep='\t')
samples <- as.character(sample_file[1,])
group <- as.character(sample_file[2,])
batch <- as.character(sample_file[6,])
libSize <- as.numeric(sample_file[3,])
y <- peaks[,-c(1,2,3)]
coords <- peaks[,c(1,2,3)]
row.names(y) <- paste(peaks[,1],peaks[,2],peaks[,3],sep='_')
inds <- rep(-1,0)
if (dim(sample_file)[1] >= 4)
inds <- c(inds,which(sample_file[4,] == 1))
inds <- c(inds,grep('input',samples))
sink(log_file_name,append = T)
print('Input samples and file removed:')
samples[sort(inds)]
sink()
samples <- samples[-inds]
group <- group[-inds]
libSize <- libSize[-inds]
batches <- batch[-inds]
# ======== edgeR differential binding analysis ===========
if (de_type == 0) {
# Combine all the experimental factors (tumour and date) into one combined factor (3.3.1 of edgeR user guide)
de<- DGEList(counts=y, group=samples,lib.size = libSize)
count_per_million=cpm(y,lib.size = libSize)
agroup <- as.character(samples)
design <- model.matrix(~0+agroup)
conds_b <- as.character(group[which(!duplicated(group))])
#conds <- as.character(samples[which(!duplicated(samples))])
conds <- as.character(colnames(design))
}
if (de_type == 1) {
# Combine all tumours in one factors (flatten date information)
de<- DGEList(counts=y, group=group,lib.size = libSize)
agroup <- as.character(group)
design <- model.matrix(~0+agroup)
conds_b <- as.character(group[which(!duplicated(group))])
conds <- conds_b
}
if (de_type == 2) {
# Model explicitely all the experimental factors (tumour and date) into a linear model (3.4.2 of edgeR user guide)
de<- DGEList(counts=y, group=group,lib.size = libSize)
de<- calcNormFactors(de, method= 'none')
Batch <- factor(batches)
agroup <- as.character(samples)
design <- model.matrix(~0+agroup+Batch)
# this case seems to generate
# Error in glmFit.default(y, design = design, dispersion = dispersion, offset = offset, :
# Design matrix not of full rank. The following coefficients not estimable:
# Batchwk -
#Tumour <- factor(group)
#design <- model.matrix(~0+Tumour+Batch)
conds_b <- as.character(group[which(!duplicated(group))])
conds <- conds_b
}
de<- estimateGLMCommonDisp(de, design)
de<- estimateGLMTagwiseDisp(de, design)
fit<- glmFit(de, design)
#colnames(design)[1:length(conds)] <- conds ##here the sorting is not necessarely right!!
#colnames(design)[1:length(conds)] <- gsub('agroup','',colnames(design)[1:length(conds)])
n <- length(unique(group))
detables <- vector("list", n)
names(detables) <- conds_b
peaks_up_down <- matrix(0, dim(y)[1],n)
colnames(peaks_up_down) <- conds_b
row.names(peaks_up_down) <- row.names(y)
sink(log_file_name,append = T)
for (i in 1:length(detables))
{
print(conds_b[i])
mycond <- conds_b[i]
# inds_yes <- grep(mycond,conds) ############## i have deleted this
# inds_no <- setdiff(seq(1:length(conds)),c(inds_yes))############## i have deleted this
inds_yes <- grep(mycond,colnames(design))
inds_no <- setdiff(seq(1:dim(design)[2]),c(inds_yes))
print(paste('Current condition being tested for dba: ',mycond,sep=''))
print(paste('Indexes belonging: ',paste(inds_yes,collapse=' '),sep=''))
conds[inds_yes]
print(paste('Indexes NOT belonging: ',paste(inds_no,collapse=' '),sep=''))
conds[inds_no]
contrasts <- rep(0,dim(design)[2])
contrasts[inds_yes] <- 1/length(inds_yes)
contrasts[inds_no] <- -1/length(inds_no)
print(contrasts)
# contrasts[inds_yes] <- 1
# contrasts[inds_no] <- -1
print(paste('Contrast vector: ',paste(format(as.numeric(contrasts),digits=3),collapse=' '),sep=''))
lrt <- glmLRT(fit,contrast=contrasts)
detable<- topTags(lrt, n= nrow(y))$table
length(which(detable$logFC>0.6 & detable$FDR<= 0.05))
detable$ID <- rownames(detable)
up <- detable$ID[which(detable$logFC >= FC & detable$FDR <= th)]
down <- detable$ID[which(detable$logFC <= -FC & detable$FDR <= th)]
peaks_up_down[which(row.names(peaks_up_down) %in% up),i] <- 1
peaks_up_down[which(row.names(peaks_up_down) %in% down),i] <- -1
detables[[i]] <- detable
}
sink()
if (save_tables)
{
for (i in 1:length(detables))
write.table(detables[[i]],paste(output_prefix,'.',names(detables[i]),'_vs_all.tab',sep=''),col.names=colnames(detables[[i]]),row.names=F,sep='\t')
peaks_up_down_df <- as.data.frame(peaks_up_down,stringsAsFactors = F)
peaks_up_down_df$ID <- row.names(peaks_up_down_df)
write.table(peaks_up_down_df,paste(output_prefix,'.peaks_table_one_vs_all.tab',sep=''),col.names=colnames(peaks_up_down_df),row.names=F,sep='\t')
save(detables,peaks_up_down,file=paste(output_prefix,'.detables_one_vs_all.RData',sep=''))
# save also bed file for peak summary
peaks_up_down_df <- as.data.frame(cbind(peaks[,1:3],peaks_up_down),stringsAsFactors=F)
write.table(peaks_up_down_df,paste(output_prefix,'.peaks_table_one_vs_all.bed',sep=''),col.names=colnames(peaks_up_down_df),row.names=F,sep='\t')
}
pdf(paste(output_prefix,'.diff_peaks_n_barplot_nonorm.pdf',sep=''),12,6)
par(mfrow=c(1,3))
bars <- rep(0,n)
for (i in 1:n)
bars[i] <- length(which(peaks_up_down[,i] == 1))
barplot(bars,names=conds_b,las=2,ylab='# sepcific up-regulated peaks')
bars2 <- rep(0,n)
for (i in 1:n)
bars2[i] <- length(which(peaks_up_down[,i] == -1))
barplot(bars2,names=conds_b,las=2,ylab='# sepcific downs-regulated peaks')
barplot(rbind(bars,bars2),names=conds_b,las=2)
legend(1,10000,c('up','down'),text.col = grey.colors(2),fill=grey.colors(2))
dev.off()
all <- (rowSums(abs(peaks_up_down)))
table(all)
inv <- row.names(peaks_up_down)[which(all == 0)]
inv_means <- rowMeans(y[which(row.names(y) %in% inv),])
noinv_means <- rowMeans(y[which(!row.names(y) %in% inv),])
pdf(paste(output_prefix,'.invariant_boxplot_nonorm.pdf',sep=''),4,8)
par(mfrow=c(2,1))
boxplot(log2(inv_means),log2(noinv_means),outline=F,names=c('invariant','variant'),ylab='all tumors mean signal (log2)')
boxplot((inv_means),(noinv_means),outline=F,names=c('invariant','variant'),ylab='all tumors mean signal')
dev.off()
# =====================================
# ===== in bash do the following: =====
# =====================================
## in bash do the following:
#
#for f in `ls *_vs_all.tab| grep -v table_one`; do cat $f | awk -F "\t" '{ if(($1 >= 0.6) && ($5 < 0.05)) { print } }' | awk -F "\t" '{print $6}'| sed 's/_/ /g' | sed 's/"//g' |awk '{print $1"\t"$2"\t"$3}' | sortBed -i - > ${f%%.tab}_up.bed; done
#for f in `ls *_vs_all.tab| grep -v table_one`; do cat $f | awk -F "\t" '{ if($1 >= 0.6 && $5 < 0.05) print}' | awk -F "\t" '{print $6}'| sed 's/_/ /g' | sed 's/"//g' |awk '{print $1"\t"$2"\t"$3}' | sortBed -i - > ${f%%.tab}_up.bed; done
# for f in `ls *_vs_all.tab| grep -v table_one`;
# do
# cat $f | awk -F "\t" '{ if(($1 >= 1.5) && ($5 < 0.05)) { print } }' | awk -F "\t" '{print $6}'| sed 's/_/ /g' | sed 's/"//g' |awk '{print $1"\t"$2"\t"$3}' | sortBed -i - > ./fc_2/${f%%.tab}.1.5_up.bed;
# done
## in bash generate bed files with also logcpm on the 4th columnfor
#for f in `ls *_vs_all.tab| grep -v table_one`; do cat $f | awk -F "\t" '{ if(($1 >= 1) && ($5 < 0.05)) { print } }' | awk -F "\t" '{print $6}'| sed 's/_/ /g' | sed 's/"//g' |awk '{print $1"\t"$2"\t"$3}' | sortBed -i - > ${f%%.tab}_up_lfc1.bed; done
#
# # =================================
# # ===== PAIR BY PAIR ANALYSIS =====
# # =================================
# # 1) for each pair (identified with naming automatically)
# # 2) do diff expr
# # 3) if genes is up write peak name in the list relative to the given condition (if it's down write in the other list)
# detables_pairs <- vector("list", (n**2 - n) / 2)
# detables_pairs_names <- rep('', (n**2 - n) / 2)
# all_up <- vector("list", n)
# all_peak_tables <- vector("list", n)
# all_up_unique <- vector("list", n)
# peaks_up_down_unique <- matrix(0, dim(y[1]),n)
# peaks_up_down_count <- matrix(0, dim(y[1]),n)
# row.names(peaks_up_down_unique) <- row.names(y)
# row.names(peaks_up_down_count) <- row.names(y)
#
# sink(log_file_name,append = T)
# count <- 1
# for (i in 1:(length(conds_b)-1)) {
# for (k in (i+1):length(conds_b)) {
# cond1 <- conds_b[i]; cond2 <- conds_b[k]
# detables_pairs_names[count] <- paste(cond1,'vs',cond2,sep='_')
# print (paste(i,"versus",k))
# inds_yes <- grep(cond1,conds); inds_no <- grep(cond2,conds)
# contrasts <- rep(0,dim(design)[2])
# contrasts[inds_yes] <- 1/length(inds_yes)
# contrasts[inds_no] <- -1/length(inds_no)
#
# print(paste('Currecnt condition being tested for dba: ',cond1,' vs ',cond2,sep=''))
# print(paste('Indexes belonging to condition_1: ',paste(inds_yes,collapse=' '),sep=''))
# print(paste('Indexes belonging to condition_2: ',paste(inds_no,collapse=' '),sep=''))
# print(paste('Contrast vector: ',paste(format(as.numeric(contrasts),digits=3),collapse=' '),sep=''))
#
# lrt <- glmLRT(fit,contrast=contrasts)
# detable<- topTags(lrt, n= nrow(y))$table
# detable$ID <- rownames(detable)
# up <- detable$ID[which(detable$logFC >= FC & detable$FDR <= th)]
# down <- detable$ID[which(detable$logFC <= -FC & detable$FDR <= th)]
# all_up[[i]] <- c(all_up[[i]],up)
# all_up[[k]] <- c(all_up[[k]],down)
# detables_pairs[[count]] <- detable
# count <- count + 1
# }
# }
# names(detables_pairs) <- detables_pairs_names
# sink()
#
# # 4) after all pairs have been done, take unique of peak name per tumor and fill similar table to peaks_up_down_n
# # 1 will mean that peak has figured has up-regulated in that condition when compared to at least another condition
# # (alternative count how many times in pair comparisons and write the number here instead on 1)
# tmp1 <- as.data.frame(peaks_up_down_count,stringsAsFactors = F)
# colnames(tmp1) <- conds_b
# tmp1$id <- row.names(peaks_up_down_count)
# for (i in 1:length(conds_b)) {
# all_up_unique[[i]] <- unique(all_up[[i]])
# all_peak_tables[[i]] <- table(all_up[[i]])
# peaks_up_down_unique[which(row.names(peaks_up_down_unique) %in% all_up_unique[[i]]),i] <- 1
# tmp2 <- as.data.frame(all_peak_tables[[i]],stringsAsFactors = F)
# colnames(tmp2) <- c('Var1',paste(conds_b[i],'_count',sep=''))
# tmp1 <- merge(tmp1,tmp2,by.x='id',by.y='Var1',all=T)
# tmp1[which(!is.na(tmp1[,n+2])),(i+1)] <- tmp1[which(!is.na(tmp1[,n+2])),n+2]
# tmp1 <- tmp1[-(n+2)]
# }
# tmp1$id <- factor(tmp1$id, levels=row.names(peaks_up_down_unique))
# output <- tmp1[order(tmp1$id),]
# peaks_up_down_count <- output[,2:(n+1)]
# row.names(peaks_up_down_count) <- output[,1]
#
# # save to file: peaks_up_down_count, peaks_up_down_unique;
#
# if (save_tables)
# {
# for (i in 1:length(detables_pairs))
# write.table(detables_pairs[[i]],paste(output_prefix,'.',names(detables_pairs[i]),'.tab',sep=''),col.names=colnames(detables_pairs[[i]]),row.names=F,sep='\t')
# peaks_up_down_count_df <- as.data.frame(peaks_up_down_count,stringsAsFactors = F)
# peaks_up_down_count_df$ID <- row.names(peaks_up_down_count_df)
# write.table(peaks_up_down_count_df,paste(output_prefix,'.peaks_table_pairs.tab',sep=''),col.names=colnames(peaks_up_down_count_df),row.names=F,sep='\t')
# save(detables_pairs,peaks_up_down_count,file=paste(output_prefix,'.detables_one_vs_one.RData',sep=''))
# colnames(peaks_up_down_unique) <- colnames(peaks_up_down_count)
# peaks_up_down_unique_df <- as.data.frame(peaks_up_down_unique,stringsAsFactors = F)
# peaks_up_down_unique_df$ID <- row.names(peaks_up_down_unique_df)
# write.table(peaks_up_down_unique_df,paste(output_prefix,'.peaks_table_pairs_unique.tab',sep=''),col.names=colnames(peaks_up_down_count_df),row.names=F,sep='\t')
# # save also bed file for peak summary
# peaks_up_down_count_df <- as.data.frame(cbind(peaks[,1:3],peaks_up_down_count),stringsAsFactors=F)
# write.table(peaks_up_down_count_df,paste(output_prefix,'.peaks_table_one_vs_one_count.bed',sep=''),col.names=colnames(peaks_up_down_count_df),row.names=F,sep='\t')
# peaks_up_down_unique_df <- as.data.frame(cbind(peaks[,1:3],peaks_up_down_unique),stringsAsFactors=F)
# write.table(peaks_up_down_unique_df,paste(output_prefix,'.peaks_table_one_vs_one_unique.bed',sep=''),col.names=colnames(peaks_up_down_unique_df),row.names=F,sep='\t')
# }
# # system('gzip *.tab')
# # system('gzip *.bed')