forked from jminnier/STARTapp
-
Notifications
You must be signed in to change notification settings - Fork 0
/
save_example_data.R
184 lines (140 loc) · 6.92 KB
/
save_example_data.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
## ==================================================================================== ##
# START Shiny App for analysis and visualization of transcriptome data.
# Copyright (C) 2016 Jessica Minnier
#
# This program is free software: you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
# the Free Software Foundation, either version 3 of the License, or
# (at your option) any later version.
#
# This program is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
# GNU General Public License for more details.
#
# You should have received a copy of the GNU General Public License
# along with this program. If not, see <http://www.gnu.org/licenses/>.
#
# You may contact the author of this code, Jessica Minnier, at <[email protected]>
## ==================================================================================== ##
##
source("helpers.R")
seqdata <- read.csv("../old_data/mousecounts_example0.csv",stringsAsFactors = FALSE)
numgeneids <- 2
#scramble real data and replace NAs with random counts
seqdatascr = seqdata
set.seed(100)
tmpind = sample(1:nrow(seqdata))
seqdatascr[,1:numgeneids] = seqdata[tmpind,1:numgeneids]
counts = seqdatascr[,-(1:numgeneids)]
tmpind = which(is.na(counts),arr.ind=T)
if(length(tmpind)>0) {
set.seed(100)
tmprepl = round(runif(nrow(tmpind),min=0,max=10))
counts[tmpind] <- tmprepl
}
seqdatascr = cbind(seqdatascr[,1:numgeneids],counts)
write.csv(seqdatascr,"data/mousecounts_example.csv",row.names = FALSE)
seqdata <- read.csv("data/mousecounts_example.csv",stringsAsFactors = FALSE)
numgeneids <- 2
sampleid <- colnames(seqdata[,-(1:numgeneids)])
tmpnames <- do.call(rbind,strsplit(sampleid,"_",fixed=TRUE))
group_names <- unique(tmpnames[,1])
group <- tmpnames[,1]
rep_id <- tmpnames[,2]
sampledata = data.frame(sampleid,group,rep_id)
print("analyzing data: input data")
countdata <- seqdata
counts <- countdata[,-(1:numgeneids),drop=FALSE]
geneids <- countdata[,1:numgeneids,drop=FALSE]
tmpkeep = which(apply(is.na(geneids),1,mean)<1) #remove rows with no gene identifiers
print(paste0("Num genes kept: ",length(tmpkeep)," of ", nrow(geneids)))
counts = counts[tmpkeep,,drop=FALSE]
geneids = geneids[tmpkeep,,drop=FALSE]
countdata = countdata[tmpkeep,,drop=FALSE]
geneids = geneids%>%unite_("unique_id",colnames(geneids),remove = FALSE)
#add in catch if geneids not unique?
#handle NAs
counts[which(is.na(counts),arr.ind=T)] <- 0 #allow choice of this or removal
rownames(counts) = geneids$unique_id
design <- model.matrix(~0+sampledata$group) # allow selection of reference group
colnames(design) = levels(as.factor(sampledata$group))
log2cpm <- cpm(counts, prior.count=0.5, log=TRUE)
#voom+limma
dge <- DGEList(counts=counts) #TMM normalization first
dge <- calcNormFactors(dge)
# v <- voom(dge,design,plot=FALSE)
v <- voom(dge,design,plot=FALSE,normalize.method = "cyclicloess")
# v <- voom(counts,design,plot=TRUE,normalize="quantile") #use this to allow different normalization
#fit <- lmFit(v,design)
#fit <- eBayes(fit)
expr_data = v$E
tmpgroup = sampledata$group
#contrasts(tmpgroup)
lmobj_res = list()
for(ii in 1:length(group_names)) {
grp = relevel(tmpgroup,ref= group_names[ii] )
lm.obj = lm(t(expr_data) ~ grp)
beta.lm = t(lm.obj$coefficients)
pval.lm = t(lm.pval(lm.obj)$pval)
pval.adj.lm = apply(pval.lm,2,p.adjust,method="BH")
colnames(beta.lm) = colnames(pval.lm) = colnames(pval.adj.lm) = gsub("grp","",colnames(beta.lm))
tmpout = cbind(melt(beta.lm[,-1,drop=FALSE]),melt(pval.lm[,-1,drop=FALSE])$value,melt(pval.adj.lm[,-1,drop=FALSE])$value)
colnames(tmpout) = c("unique_id","numer_group","logFC","P.Value","adj.P.Val")
tmpout$denom_group = group_names[ii]
tmpout$test = with(tmpout, paste(numer_group,denom_group,sep="/"))
tmpout = tmpout[,c("unique_id","test","denom_group","numer_group","logFC","P.Value","adj.P.Val")]
lmobj_res[[ii]] = tmpout
}
lmobj_res = do.call(rbind,lmobj_res)
pvals = lmobj_res%>%select(unique_id,test,adj.P.Val)%>%spread(test,adj.P.Val)
logfcs = lmobj_res%>%select(unique_id,test,logFC)%>%spread(test,logFC)
colnames(pvals)[-1] = paste0("padj_",colnames(pvals)[-1])
colnames(logfcs)[-1] = paste0("logFC_",colnames(logfcs)[-1])
tmpdat = cbind(geneids,log2cpm)
tmpdat = left_join(tmpdat,logfcs)
tmpdat = left_join(tmpdat,pvals)
data_results_table = tmpdat%>%select(-unique_id) #save this into csv
tmpexprdata = data.frame("unique_id" =geneids$unique_id,expr_data)
tmpcountdata = data.frame("unique_id"=geneids$unique_id,countdata)
tmplog2cpm = data.frame("unique_id"=geneids$unique_id,log2cpm)
log2cpm_long = melt(tmplog2cpm,variable.name = "sampleid",value.name="log2cpm")
countdata_long = melt(tmpcountdata,variable.name = "sampleid",value.name="count")
#countdata_long$log2count = log2(countdata_long$count+.25)
exprdata_long = melt(tmpexprdata,variable.name = "sampleid",value.name="log2cpm_voom")
data_long = left_join(countdata_long,log2cpm_long)
data_long = left_join(data_long,exprdata_long)
data_long$group = do.call(rbind,strsplit(as.character(data_long$sampleid),"_",fixed=TRUE))[,1]
tmpgeneidnames = colnames(geneids%>%select(-unique_id))
data_long = data_long%>%select(-one_of(tmpgeneidnames))
print('analyze data: done')
# tmpcounts = data.frame("unique_id"=geneids$unique_id,counts)
# countdata_long = melt(tmpcounts,variable.name = "sampleid",value.name="count")
# tmplog2cpm = data.frame("unique_id"=geneids$unique_id,log2cpm)
# log2cpm_long = melt(tmplog2cpm,variable.name = "sampleid",value.name="log2cpm")
# data_long = left_join(countdata_long,log2cpm_long)
#
# data_long$group = do.call(rbind,strsplit(as.character(data_long$sampleid),"_",fixed=TRUE))[,1]
# #expr_data = tmplog2cpm[,-1]
#
# print('analyze data: done')
data = countdata
results = lmobj_res
#after running analysis pipeline, export this code to another example construction file
save(countdata,group_names,sampledata,results,data_long,geneids,expr_data,
file="data/mousecounts_example_analysis_results.RData")
# use for example when including p-values and fold changes
load('data/mousecounts_example_analysis_results.RData')
pvals = results%>%select(unique_id,test,adj.P.Val)%>%spread(test,adj.P.Val)
logfcs = results%>%select(unique_id,test,logFC)%>%spread(test,logFC)
colnames(pvals)[-1] = paste0("padj_",colnames(pvals)[-1])
colnames(logfcs)[-1] = paste0("logFC_",colnames(logfcs)[-1])
tmpdat = cbind(geneids,log2cpm)
tmpdat = left_join(tmpdat,logfcs)
tmpdat = left_join(tmpdat,pvals)
# For example uploads into app
example_data_results = tmpdat%>%select(-unique_id)
save(example_data_results,file="data/mousecounts_example_analyzed.RData")
write.csv(example_data_results,file="data/mousecounts_example_analyzed.csv",quote=FALSE,row.names=FALSE)
write.csv(example_data_results[1:100,],file="data/exampleanalysisres_short.csv",quote=FALSE,row.names=FALSE)
write.csv(seqdata[1:100,],"data/examplecounts_short.csv",row.names = FALSE)