forked from angieyen/ChromDiff
-
Notifications
You must be signed in to change notification settings - Fork 0
/
clust_funcs.R
207 lines (190 loc) · 8.43 KB
/
clust_funcs.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
source("setvars.R", chdir=T)
source("funcs.R", chdir=T)
## function for drawing clusters
final.drawing = function(horiz.ints, vert.ints, linewidth, minclustsize, total_rows) {
#print(horiz.ints)
#print(vert.ints)
# draw horizontal line
abline(h=horiz.ints, lwd=linewidth)
## draw rectangles for groups that meet the cutoff
drawRects(vert.ints, linewidth, minclustsize, total_rows)
}
drawRects <- function(vert.ints, linewidth, minclustsize, nrows) {
ybottom=0.5
ytop=nrows+.5
if(length(vert.ints)>0) {
for(ind in 1:length(vert.ints)) {
currclust=vert.ints[[ind]]
if((currclust[2]-currclust[1])>=minclustsize) {
rect(currclust[1]-.5, ybottom, currclust[2]+.5, ytop, lwd=linewidth)
}
}
}
}
get_vert.ints_dend = function(plotdir, mat.to.plot, noreps=FALSE) {
dendfile=paste(plotdir,"sigfeats_feats_dend_matched.rdata", sep="")
d1file=paste(plotdir, "sigfeats_feats_d1_matched.rdata", sep="")
d3file=paste(plotdir, "sigfeats_feats_d3_matched.rdata", sep="")
groupings.file=paste(plotdir,"sigfeats_feats_colname_groupings_matched.rdata", sep="")
## if no repeat genes
if(!file.exists(d1file)) {
if(!file.exists(dendfile)) {error(paste("Neither file ", d1file, " nor ", dendfile, " exist.", sep=""))}
load(dendfile)
vert.ints=get_vert_ints_height(colnames(mat.to.plot),dend, heightcutoff)
vert.ints.noreps=vert.ints
#d1offset=0
#d2offset=0
} else {
## if there were repeat genes
load(groupings.file)
sigfeats=colnames(mat.to.plot)
totalsize=length(sigfeats)
#print(sigfeats)
#print(groupings)
## Use the saved dendrogram and pvclust info if possible
if(file.exists(d1file)) {
load(d1file)
#print(d1file)
d1offset=length(final.group1.featnames)
if(length(final.group1.featnames)>1) {
d1.vert.ints=get_vert_ints_height(colnames(mat.to.plot), d1, heightcutoff)
if(d1offset!=length(labels(d1))) { warning("There should be the same length of elements in d1 dendrogram as final.group1.featnames")}
} else {
#print(length(final.group1.featnames))
## there should only be 1 feature in group 1 in this case
if(d1offset!=1) {warning("There should only be 1 feature in group 1...")}
d1.vert.ints=list(c(1,d1offset))
}
} else {
## use grouping information directly
warning("No saved dendrogram but using grouping information for group 1...")
d1offset=length(which(groupings==1))
if(d1offset>0) {
d1.vert.ints=list(c(1,d1offset))
}
}
### get d2 ints based on gene names
group2feats=names(groupings)[which(groupings==2)]
d2offset=length(group2feats)
if(d2offset>0) {
group2genes=get_genes_only(group2feats)
d2.vert.ints=list()
## figure out which genes are right before divisions (last genes in cluster)
for(ind in 1:length(d1.vert.ints)) {
dividing.genes = get_genes_only(sigfeats[d1.vert.ints[[ind]]])
dividing.inds=match_last(dividing.genes, group2genes)
d2.vert.ints[[ind]]=dividing.inds+d1offset
}
} else {
warning("Group 2 should not have a 0 or negative offset")
}
d3offset=length(which(groupings==3))
if(d3offset==0) {
d3.vert.ints=list()
} else {
if(file.exists(d3file)) {
load(d3file)
if(d3offset!=length(final.group3.featnames)){warning("Group 3 offset is not consistent...") }
if(length(final.group3.featnames)>1) {
d3.vert.ints=get_vert_ints_height(colnames(mat.to.plot), d3, heightcutoff)
} else if(d3offset==1) {
if (length(final.group3.featnames)!=1){ warning("There should be only 1 feature in group 3...")}
d3.vert.ints=list(c((d1offset+d2offset+1),(d1offset+d2offset+1)))
}
} else {
## use grouping information directly
warning("No saved dendrogram but using grouping information for group 1...")
d3.vert.ints=list(c(d1offset+d2offset+1,d1offset+d2offset+d3offset))
}
}
if(d1offset + d2offset+d3offset != totalsize) {
warning("Error with grouping sizes")
}
vert.ints=c(d1.vert.ints, d2.vert.ints, d3.vert.ints)
d3.vert.ints.noreps=lapply(d3.vert.ints, function(x){return(x-d2offset)})
vert.ints.noreps=c(d1.vert.ints, d3.vert.ints.noreps)
}
if(noreps) {
returnlist=list(vert.ints.noreps=vert.ints.noreps, mat.to.plot=mat.to.plot)
} else {
returnlist= list(vert.ints=vert.ints, mat.to.plot=mat.to.plot)
}
return(returnlist)
}
get_vert.ints_assocstate = function(plotdir, mat.to.plot) {
## to order by associated chr state
colstates = as.numeric(get_chrstates_only(colnames(mat.to.plot)))
new_order=order(colstates)
mat.to.plot=mat.to.plot[,new_order]
first.inst=sapply(min(colstates):max(colstates), function(x) {which(colstates[new_order]==x)[1]})
first.inst=first.inst[which(!is.na(first.inst))]
vert.ints=list()
for(ind in 1:length(first.inst)) {
if(ind<length(first.inst)) {
vert.ints[[ind]]=c(first.inst[ind], first.inst[ind+1]-1)
} else {
vert.ints[[ind]]=c(first.inst[ind], ncol(mat.to.plot))
}
}
return(list(vert.ints=vert.ints, mat.to.plot=mat.to.plot))
}
get_vert.ints_domstate = function(plotdir, mat.to.plot) {
## rename matrix columns so that they are gene names instead of feature ids
feat2genes=get_genes_only(colnames(mat.to.plot))
names(feat2genes)=colnames(mat.to.plot)
colnames(mat.to.plot)=feat2genes
## load in gene order from previous sig majority plotting
suffix="_matched"
plottypesuffix="_domstate"
geneorderfile=paste(plotdir, "sig_maj_geneorder", suffix, plottypesuffix, ".txt", sep="")
geneorder=as.vector(t(read.table(geneorderfile, header=TRUE)))
## reorder matrix based on previous sig majority plotting for domstate
mat.to.plot=mat.to.plot[, geneorder]
## load previously calculated dendrogram
majdendfile=paste(plotdir, "sig_maj_plot_dend", plottypesuffix, ".rdata", sep="")
load(majdendfile)
## get clusterings based on previous dendrogram
vert.ints=reorder_get_vert_ints_height(colnames(mat.to.plot), dend, heightcutoff)
## convert the colnames back to features
genes2feats=names(feat2genes)
names(genes2feats)=feat2genes
colnames(mat.to.plot)=genes2feats[colnames(mat.to.plot)]
return(list(vert.ints=vert.ints, mat.to.plot=mat.to.plot))
}
get_vert.ints.file=function(plotdir, plottypesuffix) {
vert.ints.file=paste(plotdir, "sigfeats_feats_clust_vert_ints", plottypesuffix, ".rdata", sep="")
return(vert.ints.file)
}
get_vert.ints_combinations = function(plotdir, mat.to.plot) {
sigpvals.order.file=paste(plotdir, "sig.pvals.rdata", sep="")
load(sigpvals.order.file)
## rename matrix columns so that they are gene names instead of feature ids
feat2genes=get_genes_only(colnames(mat.to.plot))
names(feat2genes)=colnames(mat.to.plot)
colnames(mat.to.plot)=feat2genes
## reorder based on previous ordering
mat.to.plot=mat.to.plot[, sigpvals.order]
###FIX THIS###
#Don't have clusterings because don't have dendrogram (only have ordering)
#dend=sigpvals.dend
#vert.ints=reorder_get_vert_ints_height(colnames(mat.to.plot), dend, heightcutoff)
vert.ints=list()
## convert the colnames back to features
genes2feats=names(feat2genes)
names(genes2feats)=feat2genes
colnames(mat.to.plot)=genes2feats[colnames(mat.to.plot)]
return(list(vert.ints=vert.ints, mat.to.plot=mat.to.plot))
}
get_vert.ints = function(plottype.str, plotdir, mat.to.plot) {
list[reorder_by_assocstate, reorder_by_dend, reorder_by_domstate, reorder_by_combinations] = get_plottype_bools(plottype.str)
if(reorder_by_dend) {
ls= get_vert.ints_dend(plotdir, mat.to.plot)
} else if(reorder_by_assocstate){
ls=get_vert.ints_assocstate(plotdir, mat.to.plot)
} else if(reorder_by_domstate) {
ls=get_vert.ints_domstate(plotdir, mat.to.plot)
} else if(reorder_by_combinations) {
ls= get_vert.ints_combinations(plotdir, mat.to.plot)
}
return(ls)
}