-
Notifications
You must be signed in to change notification settings - Fork 0
/
add_expression_data_VST_aml_TCGA.R
212 lines (151 loc) · 10.7 KB
/
add_expression_data_VST_aml_TCGA.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
### reattemt to normalize log2fc values
## get log2-fold expression values of T-ALL samples
library(DESeq2)
countsdir <- "/srv/shared/vanloo/home/jdemeul/projects/2016_mansour_ASE_T-ALL/results/AML_TCGA/htseq_counts/"
annotsfile <- "/srv/shared/vanloo/home/jdemeul/projects/2016_mansour_ASE_T-ALL/results/AML_TCGA/annotations/gdc_sample_sheet.2018-07-30-2.tsv"
# setwd(countsdir)
annots <- read.delim(file = annotsfile, as.is = T)
sampleFiles <- sub(pattern = ".gz$", replacement = "", x = annots$File.Name)
# list.files(path = countsdir, pattern = ".htseq.counts")
sampleName <- annots$Sample.ID
sampleTable <- data.frame(sampleName = sampleName,
fileName = sampleFiles,
condition = "T-ALL")
ddsHTSeq <- DESeqDataSetFromHTSeqCount(sampleTable = sampleTable,
directory = countsdir,
design= ~ 1)
# ddsHTSeq <- ddsHTSeq[ rowSums(counts(ddsHTSeq)) >= 100, ]
dds <- estimateSizeFactors(ddsHTSeq)
dds_vst <- vst(dds, blind = T)
# head(assay(dds_vst), 3)
# library(vsn)
# meanSdPlot(assay(dds_vst))
# ntd <- normTransform(dds)
# meanSdPlot(assay(ntd))
# hist(assay(dds_vst)[3,])
dds_vstmeans <- rowMeans(x = assay(dds_vst))
vst_fc <- assay(dds_vst) - dds_vstmeans
plot(dds_vstmeans, vst_fc[,1])
plot(log10(rowMeans(counts(dds, normalized=TRUE))+1), vst_fc[,1])
resdf <- as.data.frame(counts(dds, normalized=TRUE))
l2fcdf <- as.data.frame(vst_fc)
gene_ids_names <- read.delim(file = "/srv/shared/vanloo/home/jdemeul/projects/2016_mansour_ASE_T-ALL/results/20180525_HTSeqCount_gene_names.txt", as.is = T, header = F)
rownames(gene_ids_names) <- substr(gene_ids_names$V1, 1, 15)
l2fcdf$gene_name <- gene_ids_names[substr(rownames(l2fcdf), 1, 15), "V2"]
l2fcdf$mean_expression <- 2^rowMeans(x = log2(resdf+1))
resdf$gene_name <- gene_ids_names[substr(rownames(resdf), 1, 15), "V2"]
write.table(x = l2fcdf, file = "/srv/shared/vanloo/home/jdemeul/projects/2016_mansour_ASE_T-ALL/results/AML_TCGA/20180731_RNAlog2fc_vst_AML_TCGA.txt", quote = F, sep = "\t", row.names = T, col.names = T)
write.table(x = as.data.frame(assay(dds_vst)), file = "/srv/shared/vanloo/home/jdemeul/projects/2016_mansour_ASE_T-ALL/results/AML_TCGA/20180731_RNAcounts_vst_AML_TCGA.txt", quote = F, sep = "\t", row.names = T, col.names = T)
write.table(x = resdf, file = "/srv/shared/vanloo/home/jdemeul/projects/2016_mansour_ASE_T-ALL/results/AML_TCGA/20180731_RNAcounts_normalised_AML_TCGA.txt", quote = F, sep = "\t", row.names = T, col.names = T)
### show distrib of TCGA expression data
alloccgeq2 <- unique(read.delim(file = "/srv/shared/vanloo/home/jdemeul/projects/2016_mansour_ASE_T-ALL/results/20180706_l2fc_vst_AIrecurrenceGreaterThan1_table_AML.txt", as.is = T)$gene_name)
l2fcgenesofinterest <- l2fcdf[l2fcdf$gene_name %in% c(alloccgeq2, "PRDM16"), ]
library(reshape2)
library(ggplot2)
l2fcgenesofinterestmelt <- melt(data = l2fcgenesofinterest[, !colnames(l2fcgenesofinterest) == "mean_expression"], id.vars = "gene_name")
p1 <- ggplot(data = l2fcgenesofinterestmelt, mapping = aes(x = gene_name, y = value))
# p1 <- p1 + geom_point(alpha = .5, shape = 16, size = 1.5, stroke = 0)
p1 <- p1 + geom_jitter(colour = "grey", alpha = .6)
p1 <- p1 + geom_violin(scale = "width", alpha =.5)
p1 <- p1 + theme_minimal() + theme(axis.text.x = element_text(angle = 90)) + labs(y = "log2 fold change in TCGA", x = "gene name")
p1
ggsave(filename = "/srv/shared/vanloo/home/jdemeul/projects/2016_mansour_ASE_T-ALL/results/20180731_l2fc_vst_AIrecurrenceGreaterThan1_AML_TCGA.png", plot = p1, dpi = 300, width = 15, height = 6)
write.table(x = l2fcgenesofinterest, file = "/srv/shared/vanloo/home/jdemeul/projects/2016_mansour_ASE_T-ALL/results/20180731_l2fc_vst_AIrecurrenceGreaterThan1_table_AML_TCGA.txt", sep = "\t", col.names = T, row.names = F, quote = F)
#### GATA2, MECOM, PRDM16, NPM1 expression
l2fcgataprdm <- as.data.frame(t(l2fcdf[l2fcdf$gene_name %in% c("GATA2", "MECOM", "PRDM16", "NPM1"), 1:145]))
colnames(l2fcgataprdm) <- gene_ids_names[substr(colnames(l2fcgataprdm), 1, 15), "V2"]
# l2fcgataprdmmelt <- melt(data = l2fcgataprdm)
par(mfrow = c(1,3))
plot(l2fcgataprdm$PRDM16, l2fcgataprdm$GATA2, xlab = "PRDM16", ylab = "GATA2", main = "log2 fold change")
plot(l2fcgataprdm$PRDM16, l2fcgataprdm$MECOM, xlab = "PRDM16", ylab = "MECOM", main = "log2 fold change")
plot(l2fcgataprdm$GATA2, l2fcgataprdm$MECOM, xlab = "GATA2", ylab = "MECOM", main = "log2 fold change")
###### actual use of the TCGA ASE output
## combine p-values (of powered SNP loci) per gene
library(GenomicFeatures)
library(ggplot2)
### functions
combine_pvals <- function(ase_pergene) {
# ase_pergene <- ase_results_annot[4:14, ]
outdf <- data.frame(contig = ase_pergene[1, "contig"], positions = paste0(unique(ase_pergene$position), collapse = ","), pcombined = 1, gene = ase_pergene[1, "gene"], stringsAsFactors = F, power = T)
is_duplicated <- duplicated(ase_pergene$position)
has_power <- ase_pergene$filter <= 0.01
ase_pergene <- ase_pergene[!is_duplicated & has_power, ]
if (nrow(ase_pergene) > 1) {
outdf$pcombined <- fishersMethod(x = ase_pergene$pval)
} else if (nrow(ase_pergene) == 1) {
outdf$pcombined <- ase_pergene$pval
} else {
outdf$power <- F
}
return(outdf)
}
fishersMethod <- function(x) {
pchisq(q = -2 * sum(log(x)), df = 2*length(x), lower.tail = F)
}
plot_imbalance_expression <- function(imbalancedf) {
imbalancedf <- imbalancedf[order(imbalancedf$mean_expression, decreasing = F), ]
labeldf <- data.frame(pos = unlist(lapply(X = 10^(0:4), FUN = function(x) sum(imbalancedf$mean_expression < x))), expr = 10^(0:4), stringsAsFactors = F)
outdf_bak <- imbalancedf
imbalancedf <- imbalancedf[!grepl(pattern = "^HLA.*", x = imbalancedf$gene_name, perl = T) &
!grepl(pattern = "^IG[HLK].*", x = imbalancedf$gene_name, perl = T) &
!grepl(pattern = "^TR[ABDG][VCDJ].*", x = imbalancedf$gene_name, perl = T), ]
imbalancedf$notes <- ifelse(imbalancedf$padj > 0.05, "nonsig",
ifelse(imbalancedf$log2fc >= 1, "up",
ifelse(imbalancedf$log2fc <= -.73, "down", "nonsig")))
p1 <- ggplot(data = imbalancedf, mapping = aes(x = 1:nrow(imbalancedf), y = -sign(log2fc)*log10(pcombined)))
p1 <- p1 + geom_point(mapping = aes(colour = notes, size = abs(log2fc)),
show.legend = F, alpha = .4)
p1 <- p1 + geom_hline(yintercept = c(-1,1)*-log10(max(imbalancedf[imbalancedf$padj < .05, "pcombined"])), linetype = "dashed", colour = "grey") +
geom_text(data = imbalancedf[imbalancedf$notes != "nonsig", ], mapping = aes(x = which(imbalancedf$notes != "nonsig"), y = -sign(log2fc)*log10(pcombined), label = gene_name), size = 1.5, angle = 45, hjust = 0, nudge_x = nrow(imbalancedf)/250, nudge_y = 0.1, alpha = .5, show.legend = F)
p1 <- p1 + scale_y_continuous(breaks = seq(-10,10,2), oob = scales::squish, limits = c(-10,10))
p1 <- p1 + scale_x_continuous(breaks = labeldf$pos, labels = labeldf$expr, name = "mean expression (normalised)")
# p1 <- p1 + scale_color_brewer(type = "div", palette = "RdBu", direction = -1)
p1 <- p1 + scale_color_manual(values = c(nonsig = "#e0e0e0", up = "#ef8a62", down = "#67a9cf"))
p1 <- p1 + scale_size_continuous(range = c(1,7.5))
p1 <- p1 + theme_minimal() + theme(panel.grid.minor.x = element_blank(), axis.text.x = element_text(angle = -90)) + labs(x = NULL)
return(p1)
}
### end functions
# generate library of all Hs exons (should be the covered regions).
gtffile <- "/srv/shared/vanloo/pipeline-files/human/references/annotation/GDC.h38/gencode.v22.annotation.gtf.gz"
hstxdb <- makeTxDbFromGFF(file = gtffile, organism = "Homo sapiens")
# seqlevels(hstxdb) <- sub(pattern = "chr", replacement = "", x = seqlevels(seqinfo(hstxdb)))
hsexondb <- exons(x = hstxdb, columns = c("gene_id"))
seqlevelsStyle(hsexondb) <- "Ensembl"
annotsfile <- "/srv/shared/vanloo/home/jdemeul/projects/2016_mansour_ASE_T-ALL/results/AML_TCGA/annotations/gdc_sample_sheet.2018-07-30-2.tsv"
annots <- read.delim(file = annotsfile, as.is = T)
# add in the log2-fold change data and actual gene names
l2fcfile <- "/srv/shared/vanloo/home/jdemeul/projects/2016_mansour_ASE_T-ALL/results/AML_TCGA/20180731_RNAlog2fc_vst_AML_TCGA.txt"
l2fcdf <- read.delim(file = l2fcfile, as.is = T)
for (SAMPLEID in annots$Sample.ID) {
# i <- 1
# SAMPLEID <- sampledf[i, "Number"]
## read a results file
# TWESID <- paste0("WES_", sampledf[i, "t_wes_id"])
ase_resultsfile <- paste0("/srv/shared/vanloo/home/jdemeul/projects/2016_mansour_ASE_T-ALL/results/TCGA_ASE/", SAMPLEID, "/", SAMPLEID, "_ase_out.txt")
if (!file.exists(ase_resultsfile))
next
ase_results <- read.delim(file = ase_resultsfile, as.is = T)
# make results into GRanges object, identify all exonic SNPs and create new df with all of these (contains duplicate SNPs)
asegr <- GRanges(seqnames = ase_results$contig, ranges = IRanges(start = ase_results$position, end = ase_results$position))
annothits <- findOverlaps(query = asegr, subject = hsexondb)
# in one case, there were two genes using the same exon ... this just takes the first
hitgenes <- sapply(mcols(hsexondb[subjectHits(annothits)])$gene_id, FUN = function(x) x[[1]])
ase_results_annot <- data.frame(ase_results[queryHits(annothits), colnames(ase_results) != "gene"], gene = hitgenes, stringsAsFactors = F)
# create output dataframe with combined p-value per gene + adjust for multiple testing
outdf <- do.call(rbind, by(data = ase_results_annot, INDICES = ase_results_annot$gene, FUN = combine_pvals))
outdf$padj <- 1
outdf[outdf$power, "padj"] <- p.adjust(p = outdf[outdf$power, "pcombined"], method = "fdr")
# outdf$gene_name <- gene_ids_names[outdf$gene, "V2"]
# add in the log2-fold change data and actual gene names
outdf[, c("log2fc", "mean_expression", "gene_name")] <- l2fcdf[outdf$gene, c(gsub(pattern = "-", replacement = ".", x = SAMPLEID), "mean_expression", "gene_name")]
# format
outdf$contig <- factor(outdf$contig, levels = c(1:22, "X"))
outdf <- outdf[order(outdf$contig, as.integer(unlist(lapply(strsplit(outdf$positions, split = ","), FUN = function(x) x[1])))), ]
outfile <- paste0("/srv/shared/vanloo/home/jdemeul/projects/2016_mansour_ASE_T-ALL/results/TCGA_ASE/", SAMPLEID, "/", SAMPLEID, "_imbalance_expression_vst.txt")
write.table(x = outdf, file = outfile, quote = F, sep = "\t", row.names = F, col.names = T)
# plot
p1 <- plot_imbalance_expression(imbalancedf = outdf)
plotfile <- paste0("/srv/shared/vanloo/home/jdemeul/projects/2016_mansour_ASE_T-ALL/results/TCGA_ASE/", SAMPLEID, "/", SAMPLEID, "_imbalance_expression_vst.png")
ggsave(filename = plotfile, plot = p1, dpi = 300, width = 15, height = 6)
}