-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy path7_circRNA_ORF.R
227 lines (197 loc) · 10.5 KB
/
7_circRNA_ORF.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
#### read the .fa file into the desiable format:
library(Biostrings)
faFile <- readDNAStringSet("Desktop/m6A_model_validataion_data/circRNA/circRNADb/circRNA_fasta")
seq_name <- names(faFile);
sequence <- paste(faFile);
circRNAs <- data.frame(seq_name, sequence);
#### install package for ORF predictor:
source("https://bioconductor.org/biocLite.R")
biocLite("systemPipeR")
library(systemPipeR)
#### predict the ORF for each circRNA:
circRNA_ORF=predORF(faFile, n = 1, type = "df", strand = "sense") #### circRNAs with no ORFs are not included in circRNA_ORF
write.csv(circRNA_ORF,"Desktop/m6A_model_validataion_data/circRNA/circRNA_ORF.csv",row.names = FALSE)
circRNA_ORF = read.csv("Desktop/m6A_model_validataion_data/circRNA/circRNA_ORF.csv",header = TRUE)
#### find all the DRACH sites in each circRNA
circRNAs$DRACH_all = NA;
i = 1;
while ( i <= nrow(circRNAs))
{
circRNAs$DRACH_all[i] = gregexpr("[AGT][AG]AC[ACT]", circRNAs$sequence[i]);
i = i + 1;
}
#### put each DRACH site in different rows:
circRNAsDRACHsites <- circRNAs;
circRNAsDRACHsites <- circRNAsDRACHsites[,-3];
circRNAsDRACHsites$DRACH_site = NA;
i = 1; # circRNA row
n = 1; # DRACH site row
while ( i <= nrow(circRNAs))
{
j = 1; # DRACH in each lncRNA
all_DRACH = unlist(circRNAs$DRACH_all[i]);
while ( j <= length(all_DRACH))
{
circRNAsDRACHsites[n,c(1,2)] = circRNAs[i,c(1,2)];
circRNAsDRACHsites[n,3] = all_DRACH[j];
n = n + 1;
j = j + 1;
}
i = i + 1;
}
#### find the Adenosine site
circRNAsDRACHsites$A_site = NA;
circRNAsDRACHsites$A_site = circRNAsDRACHsites$DRACH_site+2;
write.csv(circRNAsDRACHsites,"Desktop/m6A_model_validataion_data/circRNA/all_DRACH_sites_in_all_circRNAs_nofilterd.csv",row.names = FALSE)
#### feature engineering:
#### remove the A site with flanking sequence <100-bp
circRNAsDRACHsites$sequence = as.character(circRNAsDRACHsites$sequence);
circRNAsDRACHsites$seq_length = nchar(circRNAsDRACHsites$sequence)
circRNAsDRACHsites = circRNAsDRACHsites[circRNAsDRACHsites$A_site>=101, ]
circRNAsDRACHsites = circRNAsDRACHsites[circRNAsDRACHsites$A_site <= (circRNAsDRACHsites$seq_length -100), ]
#### extract the 100nt*2 +1 sequences
circRNAsDRACHsites$seq_201 = subseq(circRNAsDRACHsites$sequence, circRNAsDRACHsites$A_site-100, circRNAsDRACHsites$A_site + 100)
min(nchar(circRNAsDRACHsites$seq_201))
#### check all the seq is ATGC no N:
"N" %in% circRNAsDRACHsites$seq_201
#### one-hot encoding:
circRNAsDRACHsites$encoded_seq = as.character("X");
i = 1;
while ( i <= nrow(circRNAsDRACHsites))
{
code_seq_A = gsub("A", "1000", circRNAsDRACHsites$seq_201[i]);
code_seq_T = gsub("T", "0100", code_seq_A);
code_seq_G = gsub("G", "0010", code_seq_T);
code_seq_C = gsub("C", "0001", code_seq_G);
circRNAsDRACHsites$encoded_seq[i] = as.character(code_seq_C);
i = i + 1;
}
min(nchar(circRNAsDRACHsites$encoded_seq))
write.csv(circRNAsDRACHsites,"Desktop/m6A_model_validataion_data/circRNA/filtered_circRNAsDRACHsites_one_hot_encoded.csv",row.names = FALSE)
circRNAsDRACHsites = read.csv("Desktop/m6A_model_validataion_data/circRNA/filtered_circRNAsDRACHsites_one_hot_encoded.csv",header = TRUE)
#### converted to 3D array for deep learning prediction:
circRNA_array = array(data = 0, dim = c(4,201, nrow(circRNAsDRACHsites)));
dim(circRNA_array)
circRNAsDRACHsites$encoded_seq = as.character(circRNAsDRACHsites$encoded_seq)
s = 1;
r = 1;
c = 1;
j = 1;
while (s <= nrow(circRNAsDRACHsites) )
{
c = 1;
while ( c <= 201)
{
r = 1;
while (r <= 4)
{
circRNA_array[r,c,s] = as.integer(substr(circRNAsDRACHsites$encoded_seq[s],j,j))
r = r + 1;
j = j + 1;
}
c = c + 1;
}
s = s+1;
j = 1;
}
dim(circRNA_array)
any(is.na(circRNA_array))
#### change the index order to the ones keras package can use:
circRNA_array_new = aperm(circRNA_array, c(3,2,1));
dim(circRNA_array_new)
any(is.na(circRNA_array_new))
saveRDS(circRNA_array_new,"Desktop/m6A_model_validataion_data/circRNA/circRNA_array_data.rds")
#### predict with deep learning model in python...
#### get the predicted results
circRNAs_result = read.table("Desktop/m6A_model_validataion_data/circRNA/predict_result.txt", sep = "\n")
circRNAs_result = read.csv("Desktop/m6A_model_validataion_data/circRNA/predict_result_prob.csv",header = TRUE);
#### combine with the circRNAs all feature dataframe
circRNAsDRACHsites$predict_result = NA;
circRNAsDRACHsites$predict_result = circRNAs_result;
circRNAsDRACHsites = cbind(circRNAsDRACHsites,circRNAs_result)
#### remove useless column:
circRNAsDRACHsites = circRNAsDRACHsites[,-c(6,7)]
#### combine with ORF information
circRNAsDRACHsites$ORF_start = NA;
circRNAsDRACHsites$ORF_end = NA;
circRNAsDRACHsites$ORF_length= NA;
circRNAsDRACHsites$seq_name = as.character(circRNAsDRACHsites$seq_name)
circRNA_ORF$seqnames = as.character(circRNA_ORF$seqnames)
i = 1; # circRNAs_A_sites row
while ( i <= nrow(circRNAsDRACHsites))
{
r = NA;
r = which(circRNA_ORF$seqnames == as.character(circRNAsDRACHsites$seq_name[i]))
if( length(r) > 0)
{
circRNAsDRACHsites$ORF_start[i] = circRNA_ORF$start[r];
circRNAsDRACHsites$ORF_end[i] = circRNA_ORF$end[r];
circRNAsDRACHsites$ORF_length[i] = circRNA_ORF$width[r]
}
i = i + 1
}
#### set a filter for ORF length: 150nt
#### remove rows with no ORFs:
filtered_circRNAsDRACHsites = circRNAsDRACHsites[!is.na(circRNAsDRACHsites$ORF_length), ]
write.csv(filtered_circRNAsDRACHsites,"Desktop/m6A_model_validataion_data/circRNA/filtered_circRNAsDRACHsites_withORFs.csv",row.names = FALSE);
circRNAs_150_ORF_positive = filtered_circRNAsDRACHsites[(filtered_circRNAsDRACHsites$ORF_length >= 150 & filtered_circRNAsDRACHsites[,6] == 1), ]
circRNAs_150_ORF_negative = filtered_circRNAsDRACHsites[(filtered_circRNAsDRACHsites$ORF_length >= 150 & filtered_circRNAsDRACHsites[,6] == 0), ]
circRNAs_150_ORF_positive = filtered_circRNAsDRACHsites[(filtered_circRNAsDRACHsites$ORF_length >= 150 & filtered_circRNAsDRACHsites[,7] >= 0.8), ]
####select the positive m6A before ORF_start site, then circRNAs with translatability
circRNAs_150_ORF_positive = filtered_circRNAsDRACHsites[(filtered_circRNAsDRACHsites$ORF_length >= 150 & filtered_circRNAsDRACHsites[,7] > 0.5), ]
circRNAs_150_ORF_positive_before_ORF = circRNAs_150_ORF_positive[circRNAs_150_ORF_positive$A_site < circRNAs_150_ORF_positive$ORF_start, ]
circRNAs_150_ORF_positive_before_ORF_100nt = circRNAs_150_ORF_positive_before_ORF[circRNAs_150_ORF_positive_before_ORF$A_site > (circRNAs_150_ORF_positive_before_ORF$ORF_start-100), ]
h <- hist(circRNAs_150_ORF_positive_before_ORF_100nt$V2)
plot(h,xlim=c(0.5,1.0),xlab = "Threshold of Predicted m6A",main = "circRNAs m6A of Potential Translatability", lwd=2)
circRNAs_high_translatable = circRNAs_150_ORF_positive_before_ORF_100nt[circRNAs_150_ORF_positive_before_ORF_100nt$V2>=0.9,]
circRNAs_annotation = read.table("Desktop/m6A_model_validataion_data/circRNA/circRNADb/circRNA_dataset.txt",sep="\t")
circRNAs_annotation$V1 = as.character(circRNAs_annotation$V1)
circRNAs_high_translatable$seq_name = as.character(circRNAs_high_translatable$seq_name)
circRNAs_annotation = circRNAs_annotation[circRNAs_annotation[,1] %in% circRNAs_high_translatable$seq_name, ]
ribosomal_circRNAs = read.csv("Desktop/m6A_model_validataion_data/circRNA/ribosome_associated_circRNAs.csv",header = TRUE)
ribosomal_circRNAs$NM_id = as.character(ribosomal_circRNAs$NM_id)
circRNAs_annotation$V8 = as.character(circRNAs_annotation$V8)
ribosomal_circRNAs = ribosomal_circRNAs[(ribosomal_circRNAs$NM_id %in% circRNAs_annotation$V8) & (ribosomal_circRNAs$start %in% circRNAs_annotation$V3)&(ribosomal_circRNAs$end %in% circRNAs_annotation$V4), ]
#### calculate the relative position of positive m6A with ORFs
circRNAs_150_ORF_positive = circRNAs_150_ORF_positive[,-c(2,3,6)]
circRNAs_150_ORF_negative = circRNAs_150_ORF_negative[,-c(2,3,6)]
circRNAs_150_ORF_positive = circRNAs_150_ORF_positive[,-c(2,3)]
write.csv(circRNAs_150_ORF_positive,"Desktop/m6A_model_validataion_data/circRNA/150_ORF_positive_m6A.csv",row.names = FALSE)
write.csv(circRNAs_150_ORF_positive,"Desktop/m6A_model_validataion_data/circRNA/150_ORF_positive_m6A_prob_0.8.csv",row.names = FALSE)
#### m6A site distribution in circRNAs
par(mfrow=c(2,1))
###positive
circRNAs_150_ORF_positive$A_relative_position = circRNAs_150_ORF_positive$A_site/circRNAs_150_ORF_positive$seq_length
d <- density(circRNAs_150_ORF_positive$A_relative_position)
circRNAs_150_ORF_positive$ORF_Start_relative_position = circRNAs_150_ORF_positive$ORF_start/circRNAs_150_ORF_positive$seq_length
d2 <- density(circRNAs_150_ORF_positive$ORF_Start_relative_position)
circRNAs_150_ORF_positive$ORF_End_relative_position = circRNAs_150_ORF_positive$ORF_end/circRNAs_150_ORF_positive$seq_length
d3 <- density(circRNAs_150_ORF_positive$ORF_End_relative_position)
plot(range(d$x,d2$x,d3$x),range(d$y,d2$y,d3$y), type= "n", xlab = "Relative circRNA Position", ylab="Density",
main = "circRNAs: Predicted m6A and ORF_start/_end Distribution");
lines(d, col='red', lwd =2)
lines(d2, col ='black', lwd =2 )
lines(d3, col = 'blue',lwd = 2)
#lines(d3, col = 'blue', lwd = 2)
legend(0.5, 3.0, legend = c("m6A_site","ORF_start_site","ORF_end_site"), col = c('red','black','blue'),
lty = 1:1:1, lwd = 2:2:2, cex = 1)
####negative
circRNAs_150_ORF_negative$A_relative_position = circRNAs_150_ORF_negative$A_site/circRNAs_150_ORF_negative$seq_length
d3 <- density(circRNAs_150_ORF_negative$A_relative_position)
circRNAs_150_ORF_negative$ORF_Start_relative_position = circRNAs_150_ORF_negative$ORF_start/circRNAs_150_ORF_negative$seq_length
d4 <- density(circRNAs_150_ORF_negative$ORF_Start_relative_position)
plot(range(d3$x,d4$x),range(d3$y,d4$y), type= "n", xlab = "Relative Position in circRNAs", ylab="Density",
main = "Negative_Distribution of predicted m6A sites and ORF start sites");
lines(d3, col='red', lwd=2)
lines(d4, col ='black', lwd=2 )
legend(0.5, 3.0, legend = c("m6A_site","ORF_start_site"), col = c('red','black'),
lty = 1:1, lwd = 2:2, cex = 1, box.lty = 0)
#####
circRNAs_150_ORF_positive$relative_position = NA;
circRNAs_150_ORF_positive$relative_position = (circRNAs_150_ORF_positive$A_site - circRNAs_150_ORF_positive$ORF_start + 1)/circRNAs_150_ORF_positive$ORF_length
d = density(circRNAs_150_ORF_positive$relative_position)
plot(d, xlim=c(-6,6),xlab = "Relative ORF Position",main = "circRNAs: Predicted m6A Distribution against ORF", lwd=2)
abline(v=c(0,1),col=c("blue","blue"),lty=c(2,1),lwd=c(2,2))
median(circRNAs_150_ORF_positive$relative_position)
abline(v=0.9532164,col="red",lty=2,lwd=2)
legend(2,0.2, c("ORF_start = 0","median = 0.95","ORF_end = 1"),col = c("blue","red","blue"),lty = c(2,2,1),lwd = c(2,2,2),cex = 1)