-
Notifications
You must be signed in to change notification settings - Fork 2
/
Copy pathMap2.2_Empty_Subcluster_Varimax.R
505 lines (380 loc) · 18.1 KB
/
Map2.2_Empty_Subcluster_Varimax.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
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
do_fgsea <- function(scored_genes, pathways=MSigAll, fdr=0.05, nmax=20){
res <- fgsea(pathways, scored_genes, minSize=15, maxSize=1000)
if (sum(!is.na(res$pval) & res$padj < fdr) == 0) {print("No significant enrichments"); return();}
res <- res[!is.na(res$pval) & res$padj < fdr,]
res <- res[order(res$NES),]
res_full <- res;
if (nrow(res) > nmax) {
res_pos <- data.frame(res[unlist(res$NES) >0,])
res_pos <- res_pos[!is.na(unlist(res_pos[,1])),]
res_neg <- data.frame(res[unlist(res$NES) <0,])
res_neg <- res_neg[!is.na(unlist(res_neg[,1])),]
res_pos <- res_pos[order(abs(unlist(res_pos$NES))),]
res_neg <- res_neg[order(abs(unlist(res_neg$NES))),]
res <- rbind(res_pos[1:min(nrow(res_pos), nmax),], res_neg[1:min(nrow(res_neg), nmax),])
res <- res[order(res$NES),]
}
size <- abs(res$NES)
colour <- sign(res$NES)
col_palette <- c("dodgerblue", "grey50", "firebrick")
gene_lists <- res[,"leadingEdge"]
sim_mat <- matrix(0, nrow=nrow(gene_lists), ncol=nrow(gene_lists))
for (i in 1:nrow(gene_lists)) {
for (j in i:nrow(gene_lists)) {
int <- length(intersect(unlist(gene_lists[i,1]), unlist(gene_lists[j,1])))
uni <- length(union(unlist(gene_lists[i]), unlist(gene_lists[j])))
sim_mat[i,j] <- int/uni
sim_mat[j,i] <- int/uni
colnames(sim_mat) <- unlist(res[,1])
rownames(sim_mat) <- unlist(res[,1])
}
}
require(igraph)
G <- simplify(graph_from_adjacency_matrix(sim_mat > 0.1, mode="undirected"))
plot(G, vertex.color=col_palette[colour+2], vertex.size=size*5, edge.width=2)
res$cluster <- components(G)$membership
return(list(rich=res_full, graph=G, vertex_col = col_palette[colour+2], vertex_size = size*5))
}
#### My Varimax ####
RunVarimax <- function(seur_obj, npcs=50, do.scale=FALSE, use.projected = FALSE) {
require(Seurat)
# Project pca
set.seed(1982)
if (do.scale){
obj <- ScaleData(obj, features=rownames(obj))
}
if (use.projected) {
#obj <- ProjectDim(obj, reduction="pca", do.center=TRUE, verbose=F) # - always only 5000 genes?
projected_pca <- obj@[email protected] %*% obj@[email protected]
projected_pca <- apply(projected_pca, 2, function(x){x/max(abs(x))})
} else {
projected_pca <- obj@[email protected]
}
pca_loading <- projected_pca[,1:npcs]
#pca_loading <- pca_loading/rowSums(pca_loading)
set.seed(1029)
varimax_res <- varimax(pca_loading)
rotated_pca <- obj@[email protected][,1:npcs] %*% varimax_res$rotmat
rotated_loadings <- projected_pca[,1:npcs] %*% varimax_res$rotmat
# Prevent inverted colours across components #
for (i in 1:ncol(rotated_pca)) {
# Find direction with most variability
t <- quantile(abs(rotated_pca[,i]), 0.99)
# count number of cells with extreme values of the component.
if (sum(rotated_pca[,i] < -1*t) > sum(rotated_pca[,i] > t)) {
# flip it!
rotated_pca[,i] = -1 * rotated_pca[,i]
rotated_loadings[,i] = -1 * rotated_loadings[,i]
}
}
# Rank components by variance #
varimax_var <- apply(rotated_pca, 2, var)
varimax_var <- varimax_var/sum(varimax_var)
#reorder <- order(varimax_var, decreasing=TRUE)
#rotated_pca <- rotated_pca[,reorder]
#rotated_loadings <- rotated_loadings[,reorder]
#varimax_var <- varimax_var[reorder]
# Add cell-scores to metadata for visualization #
colnames(rotated_loadings) <- paste("RotPC", 1:ncol(rotated_loadings), sep="_")
colnames(rotated_pca) <- paste("RotPC", 1:ncol(rotated_loadings), sep="_")
[email protected] <- cbind([email protected], rotated_pca[,1:npcs])
# Add Varimax as a Reduction to the object #
varimax_pca <- CreateDimReducObject(
embeddings = rotated_pca,
loadings=rotated_loadings,
key = "VM_", assay = DefaultAssay(obj), stdev=varimax_var)
obj@reductions[["varimax"]] <- varimax_pca
return(obj);
}
Interpret_varimax <- function(obj, cluster_col="Coarse_clusters", metadata_col=c("sample")) {
}
# Function to get the same colours as used by default in Seurat DimPlot
get_seurat_colours <- function(obj, group.by) {
require(scales)
identities <- [email protected][,group.by]
if (class(identities) != "factor") {
identities <- factor(identities)
}
identities <- levels(identities)
my_color_palette <- hue_pal()(length(identities))
return(my_color_palette)
#usage:
# TSNEPlot(object = object, do.return = T) +
# scale_color_manual(values = my_color_palette)
}
get_metadata_outcome <- function(obj){
tmp <- read.delim("../../Caudate_recip_data_Dec 3_20.csv", sep=",")
reject <- tmp[ match([email protected]$donor, tmp[,1]), "Post.LT.Rejection"]
reject <- factor(reject, levels=c("N", "?", "Y"))
[email protected]$trans.rejected <- reject
return(obj)
}
files <- c("AntiB_harmony_Subcluster.rds",
"Cholangiocyte_harmony_Subcluster.rds",
"Endo_harmony_Subcluster.rds",
"Hepatocyte1_harmony_Subcluster.rds",
"Hepatocyte2_harmony_Subcluster.rds",
"Macrophage_harmony_Subcluster.rds",
"NKT_harmony_Subcluster.rds",
"Stellate_harmony_Subcluster.rds")
top_up_gene_lists <- c()
top_down_gene_lists <- c()
gene_lists_names <- c();
ntop <- 50
npcs <- 50;
varimax_gene_loadings_matrix <- c();
obj_list <- list()
require(Polychrome)
donor_colourscheme <- glasbey.colors(25)
donor_colourscheme <- donor_colourscheme[-1]
for (f in files) {
require(Seurat)
set.seed(28210)
tag <- unlist(strsplit(f, "\\/"))[1]
tag <- unlist(strsplit(f, "_"))[1]
print(tag)
obj <- readRDS(f)
obj <- get_metadata_outcome(obj)
[email protected]$donor_sex <- factor([email protected]$donor_sex)
cluster_column <- "Coarse_clusters"
cluster_cols <- get_seurat_colours(obj, group.by=cluster_column)
cell_clust_col <- cluster_cols[[email protected][,cluster_column]]
print(dim(obj))
outfile <- sub("harmony", "varimax", f)
obj <- RunPCA(obj, features=rownames(obj), npcs=50)
obj <- RunVarimax(obj)
saveRDS(obj, outfile)
# cluster_col_palette <- c("#ebac23", "#b80058", "#008cf9", "#00bbad",
# "#d163e6", "#b24502", "#ff9287", "#5954d6",
# "#00c6f8", "#878500", "#00a76c", "#bdbdbd")
# cell_cluster_col <- cluster_col_palette[[email protected][,cluster_column]]
cell_loadings <- obj@[email protected]
gene_loadings <- obj@[email protected]
all_scores <- c();
all_coeffs <- c();
# ID important components.
for (i in 1:ncol(cell_loadings)) {
#sex_score <- t.test(cell_loadings[[email protected]$donor_sex=="F",i], cell_loadings[[email protected]$donor_sex=="M",i])
res <- summary(glm(cell_loadings[,i] ~ [email protected]$donor_sex + [email protected]$donor_age_group))
res2 <- summary(glm(cell_loadings[,i] ~ [email protected][,cluster_column]))
res3 <- summary(glm(cell_loadings[,i] ~ [email protected]$trans.rejected))
all_scores <- rbind(all_scores, c(res$coefficients[,4], res3$coefficients[,4], res2$coefficients[,4])) # before 4,4,1
all_coeffs <- rbind(all_coeffs, c(res$coefficients[,1], res3$coefficients[,1], res2$coefficients[,1] ))
}
rownames(all_scores) <- 1:ncol(cell_loadings)
all_scores[order(all_scores[,2]),]
rownames(all_coeffs) <- 1:ncol(cell_loadings)
all_coeffs[order(all_coeffs[,2]),]
sex_cols = c("pink", "dodgerblue")
age_cols = c("grey45", "dodgerblue", "forestgreen")
reject_cols = c("grey85", "grey65", "grey25")
varimax_association_plot <- function(component1, component2) {
layout(rbind(c(1,2,3), c(4,5,5)))
par(mar=c(4,4,1,1))
# Sex
plot(cell_loadings[,component1], cell_loadings[,component2], pch=16,
col=sex_cols[factor([email protected]$donor_sex)],
xlab=paste("RotPC", component1), ylab=paste("RotPC", component2))
legend("bottomright", levels(factor([email protected]$donor_sex)), fill=sex_cols, bty="n")
# Age
plot(cell_loadings[,component1], cell_loadings[,component2], pch=16,
col=age_cols[factor([email protected]$donor_age_group)],
xlab=paste("RotPC", component1), ylab=paste("RotPC", component2))
legend("bottomright", levels(factor([email protected]$donor_age_group)), fill=age_cols, bty="n")
# Rejection
plot(cell_loadings[,component1], cell_loadings[,component2], pch=16,
col=reject_cols[[email protected]$trans.rejected],
xlab=paste("RotPC", component1), ylab=paste("RotPC", component2))
legend("bottomright", levels([email protected]$trans.rejected), fill=reject_cols, bty="n")
# Cluster
plot(cell_loadings[,component1], cell_loadings[,component2], pch=16,
col=cell_clust_col,
xlab=paste("RotPC", component1), ylab=paste("RotPC", component2))
legend("bottomright", levels([email protected][,cluster_column]), fill=cluster_cols, bty="n")
# Donor
plot(cell_loadings[,component1], cell_loadings[,component2], pch=16,
col=donor_colourscheme[as.numeric(factor([email protected]$donor))],
xlab=paste("RotPC", component1), ylab=paste("RotPC", component2))
legend("bottomright", levels(factor([email protected]$donor)),
fill=donor_colourscheme[1:length(unique([email protected]$donor))], ncol=2, bty="n")
}
varimax_association_boxplot <- function(component, meta_column, colours) {
boxplot(cell_loadings[,component] ~ [email protected][,cluster_column][email protected][, meta_column],
col=rep(colours, each=length(unique([email protected][, cluster_column]))),
las=2, xlab="", ylab=paste("RotPC", component, sep="_"), outline=FALSE)
}
pdf(paste(tag, "associations_pval_varimax.pdf", sep="_"), width=9, height=6.5)
# Sex
sex_comp <- as.numeric(rownames(all_scores[order(all_scores[,2]),]))
varimax_association_plot(sex_comp[1], sex_comp[2])
# old
old_comp <- as.numeric(rownames(all_scores[order(all_scores[,3]),]))
varimax_association_plot(old_comp[1], old_comp[2])
# young
young_comp <- as.numeric(rownames(all_scores[order(all_scores[,4]),]))
varimax_association_plot(young_comp[1], young_comp[2])
# Rejection
reject_comp <- as.numeric(rownames(all_scores[order(all_scores[,7]),]))
varimax_association_plot(reject_comp[1], reject_comp[2])
par(mfrow=c(4,2))
par(mar=c(5,4,0,0.5))
varimax_association_boxplot(sex_comp[1], "donor_sex", sex_cols)
varimax_association_boxplot(sex_comp[2], "donor_sex", sex_cols)
varimax_association_boxplot(old_comp[1], "donor_age_group", age_cols)
varimax_association_boxplot(old_comp[2], "donor_age_group", age_cols)
varimax_association_boxplot(young_comp[1], "donor_age_group", age_cols)
varimax_association_boxplot(young_comp[2], "donor_age_group", age_cols)
varimax_association_boxplot(reject_comp[1], "trans.rejected", reject_cols)
varimax_association_boxplot(reject_comp[2], "trans.rejected", reject_cols)
dev.off()
}
# Clusters
FeaturePlot(obj, paste("RotPC_", sex_comp[1], sep=""))
tail(sort(gene_loadings[,sex_comp[1]]), 20)
require(fgsea)
immune_path <- gmtPathways("../../ExternalData/MSigDb_immune_signatures_c7.all.v7.1.symbols.gmt")
Hallmark_path <- gmtPathways("../../ExternalData/MSigDbHalmarkPathways.gmt")
MSigAll <- gmtPathways("../../ExternalData/MSigDb_curated_c2.all.v7.1.symbols.gmt")
reactome <- gmtPathways("../../ExternalData/ReactomePathways.gmt")
richments <- do_fgsea(gene_loadings[,1], Hallmark_path, fdr=0.05)
## Cluster components:
colnames(top_up_gene_lists) <- gene_lists_names
colnames(top_down_gene_lists) <- gene_lists_names
colnames(varimax_gene_loadings_matrix) <- gene_lists_names
## Pairwise similarity
set.seed(1927)
require(proxy)
intersect_similarity <- function(x, y) {length(intersect(x, y))}
cor_similarity <- function(x, y) {
keep <- !is.na(x) & !is.na(y)
cor(x[keep], y[keep])
}
pairs <- t(combn(1:(npcs*length(files)), 2))
sim_mat_int <- matrix(0, nrow=npcs*length(files), ncol=npcs*length(files))
sim_mat_cor <- matrix(0, nrow=npcs*length(files), ncol=npcs*length(files))
direction <- rep(0, nrow(pairs))
for (i in 1:nrow(pairs)) {
# Intersection
a <- intersect_similarity(top_up_gene_lists[,pairs[i,1]], top_up_gene_lists[,pairs[i,2]]) # both up
b <- intersect_similarity(top_down_gene_lists[,pairs[i,1]], top_down_gene_lists[,pairs[i,2]]) # both down
c <- intersect_similarity(top_up_gene_lists[,pairs[i,1]], top_down_gene_lists[,pairs[i,2]]) # up vs down
d <- intersect_similarity(top_down_gene_lists[,pairs[i,1]], top_up_gene_lists[,pairs[i,2]]) # down vs up
if (a+b > c+d) {
sim_mat_int[pairs[i,1], pairs[i,2]] <- a+b
direction[i] = 1
} else {
sim_mat_int[pairs[i,1], pairs[i,2]] <- c+d
direction[i] = -1
}
# correlation
s <- cor_similarity(varimax_gene_loadings_matrix[,pairs[i,1]], varimax_gene_loadings_matrix[,pairs[i,2]])
sim_mat_cor[pairs[i,1], pairs[i,2]] <- s
}
cor_loadings <- varimax_gene_loadings_matrix[rowSums(is.na(varimax_gene_loadings_matrix)) == 0,]
sim_mat_cor <- cor(cor_loadings)
diff_cor <- matrix(1, nrow=npcs*length(files), ncol=npcs*length(files))
diff_cor <- diff_cor-sim_mat_cor
#diff_cor <- diff_cor - t(sim_mat)
diag(diff_cor) <- 0
colnames(diff_cor) <- colnames(top_up_gene_lists)
rownames(diff_cor) <- colnames(top_up_gene_lists)
require("gplots")
hmap <- heatmap.2(diff_cor, scale="none", trace="none", col=colorRampPalette(c("black", "white"))(20), distfun=function(x){as.dist(x)})
diff_int <- matrix(ntop*2, nrow=npcs*length(files), ncol=npcs*length(files))
diff_int <- diff_int-sim_mat_int
diff_int <- diff_int - t(sim_mat_int)
diag(diff_int) <- 0
colnames(diff_int) <- colnames(top_up_gene_lists)
rownames(diff_int) <- colnames(top_up_gene_lists)
# MNN - no don't like this
#adj_mat <- matrix(0, nrow=nrow(diff_int), ncol=ncol(diff_int));
#for (j in 1:length(files)) {
# columns <- 1:12 + 12*(j-1)
# sub_mat <- diff_int[, columns]
# mnn <-apply(sub_mat, 1, function(x){which(x == min(x) & x < 80)[1]})
#
# adj_mat[cbind(1:nrow(adj_mat), columns[mnn])] <- 1
#}
#mnn_mat <- adj_mat * t(adj_mat)
require("gplots")
hmap <- heatmap.2(diff_int, scale="none", trace="none", col=colorRampPalette(c("black", "white"))(20), distfun=function(x){as.dist(x)})
# Get clusters
min_size = 3;
cluster_lab <- rep("", ncol(diff_int))
cluster_id = 1;
for (k in ncol(diff_int):1) {
clusters <- cutree(as.hclust(hmap$rowDendrogram), k = k)
c_size <- table(clusters)
#print(k)
for (c in names(c_size)[c_size >= min_size]) {
curr_lab <- cluster_lab[clusters == c]
if (sum(curr_lab == "") == length(curr_lab)) {
cluster_lab[clusters == c] <- cluster_id;
cluster_id <- cluster_id +1;
} else if (length(unique(curr_lab[curr_lab != ""])) == 1) {
cluster_lab[clusters == c] <- unique(cluster_lab[clusters == c & cluster_lab != ""]);
} else {
#print(curr_lab)
}
}
}
set.seed(2829)
library(RColorBrewer)
n <- max(as.numeric(cluster_lab))
qual_col_pals = brewer.pal.info[brewer.pal.info$category == 'qual',]
col_vector = unlist(mapply(brewer.pal, qual_col_pals$maxcolors, rownames(qual_col_pals)))
cluster_cols <- col_vector[as.numeric(cluster_lab)]
png("heatmap_cluster_legend.png", width=4, height=4,units="in", res=300)
tmp <- cbind(cluster_lab, cluster_cols); tmp <- unique(tmp)
pie(rep(1, nrow(tmp)), col=tmp[,2], label=tmp[,1])
dev.off()
png("heatmap_cluster.png", width=8, height=8, units="in", res=300)
hmap <- heatmap.2(diff_int, RowSideColors=cluster_cols, ColSideColors=cluster_cols, scale="none", trace="none", col=colorRampPalette(c("black", "white"))(20))
dev.off()
names(cluster_lab) <- colnames(diff_int)
saveRDS(list(varimax_gene_loadings_matrix=varimax_gene_loadings_matrix,
diff_intersection=diff_int,
diff_correlation=diff_cor,
clusters_int=cluster_lab,
clusters_cols_int=cluster_cols,
heatmap_obj_int=hmap
), file="Spatial_varimax_clustering.rds")
#
cluster_scores_mat <- c();
id = "12"
tmp <- varimax_gene_loadings_matrix[,cluster_lab == id]
cor(tmp[!is.na(rowSums(tmp)),])
names(cluster_lab)[cluster_lab == id]
tmp <- tmp[!is.na(rowSums(tmp) ),]
scores <- rowMeans(apply(tmp, 2, function(x){x/max(abs(x))}))
scores <- scores[!grepl("^RP", names(scores))]
scores <- scores[!grepl("^MT-", names(scores))]
head(sort(scores), 20)
tail(sort(scores), 20)
#require(gprofiler2)
#ge_out <- gost(names(sort(scores, decreasing=TRUE)), organism="hsapiens", ordered_query=TRUE, correction_method="fdr",
# measure_underrepresentation=TRUE, sources=c("GO:BP", "TF", "REAC", "KEGG", "WP"), custom_bg = names(scores))
#res <- ge_out$result[ge_out$result$term_size < 500 & ge_out$result$term_size > 20,]
require(fgsea)
WP_gmt <- gmtPathways(gmt.file="../ExternalData/BaderLab25Aug2020/Human_WikiPathways_August_01_2020_symbol.gmt.txt")
kegg_gmt <- gmtPathways(gmt.file="../ExternalData/BaderLab25Aug2020/Human_KEGG_August_01_2020_symbol.gmt.txt")
react_gmt <- gmtPathways(gmt.file="../ExternalData/BaderLab25Aug2020/Human_Reactome_August_01_2020_symbol.gmt.txt")
msigdb_gmt <- gmtPathways(gmt.file="../ExternalData/BaderLab25Aug2020/Human_MSigdb_August_01_2020_symbol.gmt.txt")
iob_gmt <- gmtPathways(gmt.file="../ExternalData/BaderLab25Aug2020/Human_IOB_August_01_2020_symbol.gmt.txt")
out <- fgsea(pathway=WP_gmt, stats = sort(scores, decreasing=TRUE), minSize=15, maxSize=1000); out <- out[order(out$padj),]
out2 <- fgsea(pathway=kegg_gmt, stats = sort(scores, decreasing=TRUE), minSize=15, maxSize=1000); out2 <- out2[order(out2$padj),]
out3 <- fgsea(pathway=react_gmt, stats = sort(scores, decreasing=TRUE), minSize=15, maxSize=1000); out3 <- out3[order(out3$padj),]
out4 <- fgsea(pathway=msigdb_gmt, stats = sort(scores, decreasing=TRUE), minSize=15, maxSize=1000); out4 <- out4[order(out4$padj),]
out5 <- fgsea(pathway=iob_gmt, stats = sort(scores, decreasing=TRUE), minSize=15, maxSize=1000); out5 <- out5[order(out5$padj),]
out3[out3$padj < 0.05 & out3$NES > 0,]
########################
sort(table(clusters))
colnames(gene_lists)[clusters==8]
genes <- c(unlist(gene_lists[1:50,clusters==46]))
genes1 <- c(unlist(gene_lists[1:50, colnames(gene_lists) %in% c("PSC011_4_A1_2", "PSC011_4_B1_2", "PSC011_4_C1_2", "PSC011_4_D3")]))#,
# unlist(gene_lists[51:100, colnames(gene_lists) %in% c("PSC011_4_D1_6")]))
genes2 <- c(unlist(gene_lists[51:100, colnames(gene_lists) %in% c("PSC011_4_A1_4", "PSC011_4_B1_4", "PSC011_4_C1_4")]),
unlist(gene_lists[1:50, colnames(gene_lists) %in% c("PSC011_4_D1_6")]))
### Correlation sim/diff
tmp <- varimax_gene_
require(proxy)