-
Notifications
You must be signed in to change notification settings - Fork 9
/
compute_reversed_genes.R
302 lines (246 loc) · 15.7 KB
/
compute_reversed_genes.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
#find reversed genes by other anti-cancer drugs
#t test between active and inactive drugs
#leave one-drug-out cross valdiation
library(pheatmap)
library("gplots")
library(ConsensusClusterPlus)
library(metap)
library(RColorBrewer)
cancer = "COAD"
landmark = 1
fdr_cutoff = 0.25
#CMAP score output
output_path <- paste(cancer, "/lincs_score_", landmark, ".csv", sep="")
lincs_drug_prediction = read.csv(output_path)
load("raw/lincs/lincs_signatures_cmpd_landmark.RData")
########
load(paste(cancer, '/', cancer, 'tcga_deseq_final.RData', sep='')) #
res$GeneID = sapply((res$id), function(id){unlist(strsplit(id, "\\|"))[2]})
res$symbol = sapply((res$id), function(id){unlist(strsplit(id, "\\|"))[1]})
dz_signature = subset(res, !is.na(padj) & !is.na(id) & id !='?' & padj < 1E-3 & abs(log2FoldChange) > 1.5 & abs(log2FoldChange) != Inf )
dz_sig = subset(dz_signature, select=c("GeneID", "log2FoldChange"))
##########
#visualized gene reversed
#only pick the signatures from close to the median
cell_lines = read.csv(paste("raw/cell_lines/", cancer, "_cell_lines.csv", sep=""))
#choose profiles from the same lineage
lincs_drug_prediction_subset = subset(lincs_drug_prediction, cell_id %in% cell_lines$LINCS, select=c("id", "RGES", "pert_iname", "pert_dose", "pert_time"))
###selecting median still sounds odd.
drug_cmap_score = aggregate(RGES ~ pert_iname, lincs_drug_prediction_subset, median)
drug_instances_median = merge(lincs_drug_prediction_subset, drug_cmap_score, by = c("pert_iname"))
drug_instances_median$diff = abs(drug_instances_median$RGES.x - drug_instances_median$RGES.y) #cmap_score.y is the median
drug_instances_min_diff = aggregate(diff ~ pert_iname, drug_instances_median, min)
drug_instances_select = merge(drug_instances_median, drug_instances_min_diff, by=c("pert_iname", "diff"))
drug_instances_select = drug_instances_select[!duplicated(drug_instances_select$pert_iname), ]
sig_id_selects = drug_instances_select$id
drug_dz_signature = merge(dz_sig, data.frame(GeneID = rownames(lincs_signatures), lincs_signatures[, as.character(sig_id_selects)]), by="GeneID", suffixes='')
#########################
###
#visualize the reversed gene expression
#reorder drug dz signatures
gene_ids = drug_dz_signature$GeneID
drug_dz_signature_rank = drug_dz_signature[,-1]
for (i in 1:ncol(drug_dz_signature_rank)){
drug_dz_signature_rank[,i] = rank(-1 * drug_dz_signature_rank[,i] ) #highly expressed genes ranked on the top
}
gene_ids_rank <- gene_ids[order(drug_dz_signature_rank[,1])]
drug_dz_signature_rank <- drug_dz_signature_rank[order(drug_dz_signature_rank[,1]),] #order by disease expression
lincs_drug_activity = read.csv(paste(cancer, "/lincs_drug_activity_confirmed.csv", sep=""), stringsAsFactors=F)
lincs_drug_activity = unique(subset(lincs_drug_activity, select=c("pert_iname", "doc_id", "standard_value", "standard_type", "description", "organism", "cell_line")))
lincs_drug_activity = aggregate(standard_value ~ pert_iname, lincs_drug_activity, median)
active_drug_names = lincs_drug_activity$pert_iname[lincs_drug_activity$standard_value <= 10000 & lincs_drug_activity$pert_iname %in% lincs_drug_prediction_subset$pert_iname]
inactive_drug_names = lincs_drug_activity$pert_iname[lincs_drug_activity$standard_value > 10000 & lincs_drug_activity$pert_iname %in% lincs_drug_prediction_subset$pert_iname]
col_sorted = sort(cor(drug_dz_signature_rank, method="spearman")["log2FoldChange",-1])
drug_dz_signature_rank = drug_dz_signature_rank[,c("log2FoldChange", names(col_sorted))]
drug_names = sapply(2:ncol(drug_dz_signature_rank), function(id){
lincs_drug_prediction$pert_iname[paste("X",lincs_drug_prediction$id, sep="") == names(drug_dz_signature_rank)[id]][1]
})
pdf(paste( "fig/", cancer, "_lincs_reverse_expression.pdf", sep=""))
colPal <- greenred(100)
par(mar=c(13, 6, 2, 0.5))
axiscolor = sapply(c(cancer, as.character(drug_names)), function(name){
if (name == cancer){
"black"
}else if (name %in% active_drug_names){
"red"
}else{
"green"
}
})
image(t(drug_dz_signature_rank), col=colPal, axes=F, srt=45)
axis(1, at= seq(0,1,length.out=ncol( drug_dz_signature_rank ) ), labels= FALSE)
text(x = seq(0,1,length.out=ncol( drug_dz_signature_rank ) ), c(-0.05),
labels = c( cancer,as.character(drug_names)),col=axiscolor, srt = 45, pos=2,offset=0.05, xpd = TRUE, cex=0.4)
dev.off()
##########
#unsupervise clustering to find common clusters
# results = ConsensusClusterPlus(as.matrix(drug_dz_signature_reversed),maxK=6,reps=50,pItem=0.8,pFeature=1,
# title=paste(cancer, "_cluster",sep=""),clusterAlg="hc",distance="euclidean",seed=1262118388.71279,plot="png")
# sort(results[[2]][["consensusClass"]])
#####################
#find reversed genes
###
#leave one drug out
all_drugs = unique(active_drug_names, inactive_drug_names)
down_p_value_matrix = matrix(NA, nrow=length(all_drugs), ncol=nrow(drug_dz_signature_rank))
up_p_value_matrix = matrix(NA, nrow=length(all_drugs), ncol=nrow(drug_dz_signature_rank))
for (i in 1:length(all_drugs)){
print(i)
drug_left = all_drugs[i]
active_drug_names_new = active_drug_names[!active_drug_names %in% drug_left]
inactive_drug_names_new = inactive_drug_names[!inactive_drug_names %in% drug_left]
lincs_sig_info_id = paste("X", lincs_drug_prediction$id, sep="")
active_drugs = lincs_sig_info_id[lincs_drug_prediction$pert_iname %in% active_drug_names_new & lincs_sig_info_id %in% colnames(drug_dz_signature_rank)]
inactive_drugs = lincs_sig_info_id[lincs_drug_prediction$pert_iname %in% inactive_drug_names_new & lincs_sig_info_id %in% colnames(drug_dz_signature_rank)]
down_p_value = sapply(1:nrow(drug_dz_signature_rank), function(id){
t.test(as.numeric(drug_dz_signature_rank[id, active_drugs] - drug_dz_signature_rank[id,1]) , as.numeric(drug_dz_signature_rank[id, inactive_drugs] - drug_dz_signature_rank[id,1]), alternative=c("greater"))$p.value
})
up_p_value = sapply(1:nrow(drug_dz_signature_rank), function(id){
t.test(as.numeric(drug_dz_signature_rank[id, active_drugs] - drug_dz_signature_rank[id,1]) , as.numeric(drug_dz_signature_rank[id, inactive_drugs] - drug_dz_signature_rank[id,1]), alternative=c("less"))$p.value
})
down_p_value_matrix[i,] = p.adjust(down_p_value, method="fdr") #down_p_value #
up_p_value_matrix[i,] = p.adjust(up_p_value, method="fdr") #up_p_value #
}
gene_p = data.frame(gene = gene_ids_rank, sig_down_count = apply(down_p_value_matrix, 2, function(values){sum(values<0.25)}),
sig_up_count = apply(up_p_value_matrix, 2, function(values){sum(values<0.25)}),
sig_down_meta = apply(down_p_value_matrix, 2, function(values){sumlog(values)$p}),
sig_up_meta = apply(up_p_value_matrix, 2, function(values){sumlog(values)$p})
)
gene_p_annot = merge(gene_p, res, by.x="gene", by.y="GeneID")
gene_p_annot = gene_p_annot[order(gene_p_annot$sig_down_meta), ]
gene_p_annot$sig_up_meta_padj = p.adjust(gene_p_annot$sig_up_meta, method="fdr")
gene_p_annot$sig_down_meta_padj = p.adjust(gene_p_annot$sig_down_meta, method = "fdr")
#valid_genes = data.frame(gene=gene_ids_rank, up_p=up_p_value, up_q= up_fdr, down_p = down_p_value, down_q = down_fdr)
#valid_genes = valid_genes[order(valid_genes$down_q),]
#valid_genes = subset(valid_genes, down_q < fdr_cutoff | up_q < fdr_cutoff)
#valid_genes = merge(valid_genes, res, by.x="gene", by.y="GeneID")
#write.table(valid_genes[valid_genes$up_q < fdr_cutoff,], paste(cancer,"/target_up_genes_classify_good_bad_drug.txt", sep=""), sep="\t", row.names=F, col.names=T, quote=F)
#write.table(valid_genes[valid_genes$down_q < fdr_cutoff,], paste(cancer,"/target_down_genes_classify_good_bad_drug.txt", sep=""), sep="\t", row.names=F, col.names=T, quote=F)
write.table(gene_p_annot, paste(cancer,"/target_reversed_score.txt", sep=""), sep="\t", row.names=F, col.names=T, quote=F)
#visualize only reversed geneså
###genes reversed
gene_ids = drug_dz_signature$GeneID
gene_ids_annot <- merge(gene_ids, res, by.x=1, by.y="GeneID", sort=F)
drug_dz_signature_rank = drug_dz_signature[,-1]
for (i in 1:ncol(drug_dz_signature_rank)){
drug_dz_signature_rank[,i] = rank(1 * drug_dz_signature_rank[,i] ) #highly expressed genes ranked on the top
}
drug_names = sapply(2:ncol(drug_dz_signature_rank), function(id){
lincs_drug_prediction$pert_iname[paste("X",lincs_drug_prediction$id, sep="") == as.character(names(drug_dz_signature_rank)[id])][1]
})
drug_dz_signature_reversed = (drug_dz_signature_rank - drug_dz_signature_rank[, 1])
drug_dz_signature_reversed = drug_dz_signature_reversed[, -c(1)]
rownames(drug_dz_signature_reversed) = gene_ids_annot$symbol
colnames(drug_dz_signature_reversed) = drug_names
drug_dz_signature_reversed = drug_dz_signature_reversed[rownames(drug_dz_signature_reversed) %in% as.character(gene_p_annot$symbol[gene_p_annot$sig_down_count >= length(all_drugs) | gene_p_annot$sig_up_count >= length(all_drugs)]),]
annotation = rbind(data.frame(drug = active_drug_names, type = "effective"), data.frame(drug = inactive_drug_names, type = "ineffective"))
rownames(annotation) = annotation[,1]
annotation$type = factor(annotation$type)
annotation = subset(annotation, select="type")
names(annotation) = "Activity"
Var1 = c("green", "red")
names(Var1) = c("effective", "ineffective")
ann_colors = list(Activity = c(effective = "green", ineffective="blue"))
my.cols <- greenred(100) # brewer.pal(9, "Blues")
pheatmap((drug_dz_signature_reversed), annotation = annotation, annotation_legend = T, col = my.cols,
cellheight = 12, cellwidth = 8, show_rownames = T, legend=T,
clustering_distance_cols = "euclidean", annotation_colors = ann_colors,
filename = paste( "fig/targets_", cancer, "_reversed_genes.pdf", sep="")) #
##############
#compute using rank
drug_names_no_ranked = sapply(3:ncol(drug_dz_signature), function(id){
lincs_drug_prediction$pert_iname[paste("X",lincs_drug_prediction$id, sep="") == names(drug_dz_signature_rank)[id]][1]
})
comparison_frame = drug_dz_signature[,-c(1, 2)]
gene_ids = drug_dz_signature[, 1]
sample_class = rep(0, length(drug_names_no_ranked))
sample_class[drug_names_no_ranked %in% active_drug_names] = 1
method = "SAM"
if (method == 'RANKPROD_SINGLE') {
library(RankProd)
RP_result <- RP(comparison_frame, sample_class,gene.names=gene_ids, num.perm = 100, logged = F, na.rm = TRUE, plot = FALSE, rand = 123)
# Leave logged=FALSE because topGene() converts the fold-change incorrectly!
siggenes <- topGene(RP_result,cutoff=0.1,method="pfp",logged=FALSE,gene.names=gene_ids)
# Normalize the results across methods
siggenes.result = list()
siggenes.result$UP <- (siggenes$Table1[,3:5])
colnames(siggenes.result$UP) <- c("fold.change","q.value","p.value")
siggenes.result$DOWN <- siggenes$Table2[,3:5]
colnames(siggenes.result$DOWN) <- c("fold.change","q.value","p.value")
# Since RANKPROD does the goofy condition 1 / condition 2, inverse the fold-change and convert to log if not logged.
siggenes.result$UP[,"fold.change"] <- log2(1/siggenes.result$UP[,"fold.change"])
siggenes.result$DOWN[,"fold.change"] <- log2(1/siggenes.result$DOWN[,"fold.change"])
siggenes.result$UP=data.frame(siggenes.result$UP,probe=rownames(siggenes.result$UP))
siggenes.result$DOWN=data.frame(siggenes.result$DOWN,probe=rownames(siggenes.result$DOWN))
} else {
library(siggenes)
SAM_result <- sam(comparison_frame,sample_class,rand=123,gene.names=gene_ids) #q.version=1
delta.table <- rbind(c(0,0,0),findDelta(SAM_result,fdr=0.9)) #fdr=0.9, too loose
siggenes.table <- summary(SAM_result, delta.table[ dim(delta.table)[1], 1] );
siggenes <- [email protected]
if( nrow(siggenes) > 0 ) {
siggenes.result = list()
siggenes.result$UP <- siggenes[siggenes$d.value > 0 & siggenes$rawp < 0.05,c("R.fold","q.value","rawp")]
colnames(siggenes.result$UP) <- c("fold.change","q.value","p.value")
siggenes.result$DOWN <- siggenes[siggenes$d.value < 0 & siggenes$rawp < 0.05,c("R.fold","q.value","rawp")]
colnames(siggenes.result$DOWN) <- c("fold.change","q.value","p.value")
siggenes.result$UP=data.frame(siggenes.result$UP,probe=rownames(siggenes.result$UP))
siggenes.result$DOWN=data.frame(siggenes.result$DOWN,probe=rownames(siggenes.result$DOWN))
}
genes.up <- merge(siggenes.result$UP,res, by.x="probe",by.y="GeneID",sort=F,all.x=T)
genes.down <- merge(siggenes.result$DOWN,res, by.x="probe",by.y="GeneID",sort=F,all.x=T)
}
write.table(rbind(genes.up,genes.down), paste(cancer,"/target_SAM_good_bad_drugs.txt", sep=""), sep="\t", row.names=F, col.names=T, quote=F)
#visualize
comparison_frame_subset = comparison_frame
rownames(comparison_frame_subset) = gene_ids
comparison_frame_subset = comparison_frame_subset[rownames(comparison_frame_subset) %in% c(genes.up$probe, genes.down$probe), ]
drug_names = sapply(1:ncol(comparison_frame_subset), function(id){
lincs_drug_prediction$pert_iname[paste("X",lincs_drug_prediction$id, sep="") == as.character(names(comparison_frame_subset)[id])][1]
})
rownames(comparison_frame_subset) = merge(data.frame(GeneID=rownames(comparison_frame_subset)), res, by = "GeneID", sort=F)$symbol
colnames(comparison_frame_subset) = drug_names
annotation = rbind(data.frame(drug = active_drug_names, type = "effective"), data.frame(drug = inactive_drug_names, type = "ineffective"))
rownames(annotation) = annotation[,1]
annotation$type = factor(annotation$type)
annotation = subset(annotation, select="type")
names(annotation) = "Activity"
Var1 = c("green", "red")
names(Var1) = c("effective", "ineffective")
ann_colors = list(Activity = c(effective = "green", ineffective="blue"))
my.cols <- greenred(100) # brewer.pal(9, "Blues")
pheatmap((comparison_frame_subset), annotation = annotation, annotation_legend = T, col = my.cols,
cellheight = 12, cellwidth = 8, show_rownames = T, legend=T,
clustering_distance_cols = "euclidean", annotation_colors = ann_colors,
filename = paste( "fig/targets_", cancer, "_reversed_genes_SAM.pdf", sep="")) #
#
###
#select reversed genes based on enrichment
up_genes = dz_sig$GeneID[dz_sig$GeneID %in% rownames(lincs_signatures) & dz_sig$log2FoldChange > 0]
down_genes = dz_sig$GeneID[dz_sig$GeneID %in% rownames(lincs_signatures) & dz_sig$log2FoldChange < 0]
effective_drug_signature_ids = lincs_drug_prediction[lincs_drug_prediction$pert_iname %in% active_drug_names, "id"]
ineffective_drug_signature_ids = lincs_drug_prediction[lincs_drug_prediction$pert_iname %in% inactive_drug_names, "id"]
drug_signature_rank = lincs_signatures[, as.character(sig_id_selects)]
for (i in 1:ncol(drug_signature_rank)){
drug_signature_rank[,i] = rank(-1 * drug_signature_rank[,i] ) #highly expressed genes ranked on the top
}
up_ps = NULL
for (gene in up_genes){
effective_ranks = as.numeric(drug_signature_rank[gene, colnames(drug_signature_rank) %in% effective_drug_signature_ids])
ineffective_ranks = as.numeric(drug_signature_rank[gene, colnames(drug_signature_rank) %in% ineffective_drug_signature_ids])
up_ps = c(up_ps, t.test(effective_ranks, ineffective_ranks, alternative = "greater")$p.value)
}
up_genes_p = data.frame(GeneID = up_genes, p = up_ps)
up_genes_p = merge(up_genes_p, res, by = "GeneID")
up_genes_p = up_genes_p[order(up_genes_p$p), ]
up_genes_p$p_adj = p.adjust(up_genes_p$p)
down_ps = NULL
for (gene in down_genes){
effective_ranks = as.numeric(drug_signature_rank[gene, colnames(drug_signature_rank) %in% effective_drug_signature_ids])
ineffective_ranks = as.numeric(drug_signature_rank[gene, colnames(drug_signature_rank) %in% ineffective_drug_signature_ids])
down_ps = c(down_ps, t.test(effective_ranks, ineffective_ranks, alternative = "less")$p.value)
}
down_genes_p = data.frame(GeneID = down_genes, p = down_ps)
down_genes_p = merge(down_genes_p, res, by = "GeneID")
down_genes_p = down_genes_p[order(down_genes_p$p), ]
down_genes_p$p_adj = p.adjust(down_genes_p$p)