/
SubsampleNGS.R
152 lines (128 loc) · 5.54 KB
/
SubsampleNGS.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
#documentation start
#=============================================================================
# File data
# creator: Christiane Hassenrück
# acknowledgements: Alban Ramette
# primary authority: Christiane Hassenrück
# other authorities:
#=============================================================================
# File contents
# function to repeatedly subsample Sample by OTU table to obtain more reliable alpha diversity estimates
# calculates number of OTU, chao1, inverse Simpson, absolute and relative singletons, absolute doubletons
#
# input:
# Data - sample by OTU table (samples are columns)
# n - number of iterations (how many times do you want to subsample)
# sub - library size (how many sequences per sample)
# subTable - should one iteration of rarefied OTU be writen to file (logical)
#
# output:
# iterations - list with values (number of OTU, chao1, inverse Simpson, percentage of OTUs that are
# absolute and relative singletons, and absolute doubletons) for each iteration
# summaryAlpha - mean of iterations for above indices of rarefyied data set
# summaryHill - mean of iterations for Hill numbers 0, 1, 2 of rarefyied data set
# summaryHillRaw - Hill numbers 0, 1, 2 of original data set
#
#
# dependencies:
# require(vegan)
#=============================================================================
#documentation end
SubsampleNGS <- function(Data, n, sub, subTable = F) {
if(!"vegan" %in% installed.packages()) {
install.packages("vegan")
}
require(vegan)
#removing samples with less than sub sequences
Data <- Data[, colSums(Data) >= sub]
output <- list(iterations = list(nOTU = matrix(NA, n, ncol(Data)),
chao1 = matrix(NA, n, ncol(Data)),
ace = matrix(NA, n, ncol(Data)),
invS = matrix(NA, n, ncol(Data)),
shannon = matrix(NA, n, ncol(Data)),
SSOabs = matrix(NA, n, ncol(Data)),
SSOrel = matrix(NA, n, ncol(Data)),
DSOabs = matrix(NA, n, ncol(Data))),
summaryAlpha = matrix(NA, 8, ncol(Data)),
summaryHill = matrix(NA, 3, ncol(Data)),
summaryHillRaw = matrix(NA, 3, ncol(Data))
)
colnames(output$iterations$nOTU) =
colnames(output$iterations$chao1) =
colnames(output$iterations$ace) =
colnames(output$iterations$invS) =
colnames(output$iterations$shannon) =
colnames(output$iterations$SSOabs) =
colnames(output$iterations$SSOrel) =
colnames(output$iterations$DSOabs) =
colnames(output$summaryAlpha) =
colnames(output$summaryHill) =
colnames(output$summaryHillRaw) =
colnames(Data)
rownames(output$summaryAlpha) <- names(output$iterations)
rownames(output$summaryHill) <- c("Hill0","Hill1","Hill2")
rownames(output$summaryHillRaw) <- c("Hill0","Hill1","Hill2")
for(j in 1:n){
print(j)
#subsampling
SampleOTU <- t(rrarefy(t(Data), sub))
#exclude empty OTUs
Data_new <- SampleOTU[rowSums(SampleOTU) > 0, ]
if (subTable == T & j == 1) {
write.table(Data_new, "subTable.txt", quote = F, sep = "\t")
}
#nOTU
Data01 <- Data_new
Data01[Data01 > 0] <- 1
nOTU <- apply(Data01, 2, sum)
output$iterations$nOTU[j, ] <- nOTU
#chao1 + ACE
chao1 <- estimateR(t(Data_new))
output$iterations$chao1[j, ] <- chao1[2, ]
output$iterations$ace[j, ] <- chao1[4, ]
#invS + shannon
invS <- diversity(t(Data_new), "inv")
output$iterations$invS[j, ] <- invS
Shannon <- diversity(t(Data_new), "shannon")
output$iterations$shannon[j, ] <- Shannon
#SSO
D <- t(Data_new)
if (sum(apply(D, 2, sum) == 1) >= 1) {
SSOabs.D <- data.frame(D[, apply(D, 2, sum) == 1])
SSOabsPerSample <- apply(SSOabs.D, 1, sum)
} else {
SSOabsPerSample <- rep(0, nrow(D))
}
output$iterations$SSOabs[j, ] <- round((SSOabsPerSample / nOTU) * 100, 2)
#DSOabs
if (sum(apply(D, 2, sum) == 2) >= 1) {
DSOabs.D <- data.frame(D[, apply(D, 2, sum) == 2])
DSOabsPerSample <- apply(DSOabs.D, 1, sum)
} else {
DSOabsPerSample <- rep(0, nrow(D))
}
output$iterations$DSOabs[j, ] <- round((DSOabsPerSample / nOTU) * 100, 2)
#SSOrel
# removing SSOabs
D1 <- D[, apply(D, 2, sum) > 1]
# SSOrel should have at least a sample with one sequence alone. See definition.
if (sum(apply(D1, 2, function(x) any(x == 1)) == TRUE) >= 1) {
SSOrel.D <- data.frame(D1[, apply(D1, 2, function(x) any(x == 1)) == TRUE])
# SSOrel.D contains the subtable with all SSOrel of the dataset
SSOrelPerSample <- apply(SSOrel.D, 1, function(x) sum(x == 1))
} else {
SSOrelPerSample <- rep(0, nrow(D))
}
output$iterations$SSOrel[j, ] <- round((SSOrelPerSample / nOTU) * 100, 2)
}
for (i in 1:8) {
output$summaryAlpha[i,] <- apply(output$iterations[[i]], 2, mean)
}
output$summaryHill[1, ] <- output$summaryAlpha["nOTU",]
output$summaryHill[2, ] <- exp(output$summaryAlpha["shannon",])
output$summaryHill[3, ] <- output$summaryAlpha["invS",]
output$summaryHillRaw[1, ] <- colSums(decostand(Data, method = "pa"))
output$summaryHillRaw[2, ] <- exp(diversity(t(Data), "shannon"))
output$summaryHillRaw[3, ] <- diversity(t(Data), "inv")
return(output)
}