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
|
### =========================================================================
### genotypeToSnpMatrix methods
### =========================================================================
## Coding for snpMatrix :
## 0 = missing OR multiallelic OR multi-ALT values
## 1 = homozygous reference (0|0 or 0/0)
## 2 = heterozygous (0|1 or 0/1 or 1|0 or 1/0)
## 3 = homozygous alternate (risk) allele (1|1 or 1/1)
## empty matrix to return if conditions not met
.emptySnpMatrix <- function() {
list(genotype=new("SnpMatrix"),
map=DataFrame(snp.names=character(),
allele.1=DNAStringSet(),
allele.2=DNAStringSetList(),
ignore=character()))
}
setMethod("genotypeToSnpMatrix", "CollapsedVCF",
function(x, uncertain=FALSE, ...)
{
ok <- suppressWarnings(require("snpStats", quietly=TRUE,
character.only=TRUE))
ok || stop("'snpStats' required; try biocLite('snpStats')", call.=FALSE)
alt <- alt(x)
if (is(alt, "CompressedCharacterList")) {
alt <- .toDNAStringSetList(alt)
if (all(elementLengths(alt) == 0L)) {
warning("No nucleotide ALT values were detected.")
return(.emptySnpMatrix())
}
}
ref <- ref(x)
if (ncol(x) == 0) {
warning("no samples in VCF")
}
if (!uncertain) {
gt <- geno(x)$GT
} else {
geno.cols <- row.names(geno(exptData(x)[["header"]]))
if ("GP" %in% geno.cols) {
gt <- geno(x)$GP
if (mode(gt) == "list") {
gt <- .matrixOfListsToArray(gt)
}
} else if ("GL" %in% geno.cols) {
gt <- geno(x)$GL
if (mode(gt) == "list") {
gt <- .matrixOfListsToArray(gt)
}
gt <- GLtoGP(gt)
} else {
warning("uncertain=TRUE requires GP or GL; returning NULL")
return(.emptySnpMatrix())
}
}
callGeneric(gt, ref, alt)
})
setMethod("genotypeToSnpMatrix", "array",
function(x, ref, alt, ...)
{
if (!is(ref, "DNAStringSet"))
stop("'ref' must be a DNAStringSet")
if (!is(alt, "DNAStringSetList"))
stop("'alt' must be a DNAStringSetList")
# query ref and alt alleles for valid SNPs
altelt <- elementLengths(alt) == 1L
snv <- .testForSNV(ref, alt)
# if x is a matrix, we have GT with a single value for each snp
if (is.matrix(x)) {
if (!all(altelt)) {
warning("variants with >1 ALT allele are set to NA")
x[!altelt,] <- ".|."
}
if (!all(snv)) {
warning("non-single nucleotide variations are set to NA")
x[!snv,] <- ".|."
}
map <- .genotypeToIntegerSNV(TRUE)
diploid <- x %in% names(map)
if (!all(diploid)) {
warning("non-diploid variants are set to NA")
x[!diploid] <- ".|."
}
mat <- matrix(map[x], nrow=ncol(x), ncol=nrow(x),
byrow=TRUE, dimnames=rev(dimnames(x)))
genotypes <- new("SnpMatrix", mat)
} else {
# if x is a 3D array, we have GP with multiple values for each snp
if (!all(altelt)) {
warning("variants with >1 ALT allele are set to NA")
x[!altelt,,] <- NA
}
if (!all(snv)) {
warning("non-single nucleotide variations are set to NA")
x[!snv,,] <- NA
}
# if there is more than one ALT allele for any variant,
# the 3rd dimension of the array will be too big
# any values here should already have been set to NA above
if (dim(x)[3] > 3) {
x <- x[,,1:3]
}
# for each sample, call probabilityToSnpMatrix
smlist <- list()
for (s in 1:ncol(x)) {
sm <- probabilityToSnpMatrix(x[,s,])
rownames(sm) <- colnames(x)[s]
smlist[[s]] <- sm
}
genotypes <- do.call(rbind, smlist)
}
flt <- !(snv & altelt)
map <- .createMap(rownames(x), ref, alt, flt)
list(genotypes = genotypes, map = map)
})
.createMap <- function(nms, ref, alt, flt)
{
if (is.null(ref))
DataFrame(snp.names=character(0),
allele.1=DNAStringSet(),
allele.2=DNAStringSetList(),
ignore=logical())
else
DataFrame(snp.names=nms,
allele.1=ref,
allele.2=alt,
ignore=flt)
}
probabilityToSnpMatrix <- function(probs) {
ok <- suppressWarnings(require("snpStats", quietly=TRUE,
character.only=TRUE))
ok || stop("'snpStats' required; try biocLite('snpStats')", call.=FALSE)
if (ncol(probs) != 3)
stop("input matrix should have 3 columns: P(A/A), P(A/B), P(B/B)")
# skip missing values when checking for validity of probabilities
missing <- rowSums(is.na(probs)) > 0
if (!isTRUE(all.equal(rowSums(probs[!missing,,drop=FALSE]),
rep(1,sum(!missing)),
check.attributes=FALSE,
check.names=FALSE)))
stop("sum of probabilities in each row of input matrix should = 1")
# post2g can't handle missing data
if (sum(missing) > 0) {
probs[missing,] <- 0
g <- post2g(probs)
g[missing] <- as.raw(0)
} else {
g <- post2g(probs)
}
g <- matrix(g, nrow=1, dimnames=list(NULL, rownames(probs)))
new("SnpMatrix", g)
}
GLtoGP <- function(gl) {
if (is.matrix(gl) & mode(gl) == "list") {
gp <- gl
for (i in 1:length(gp)) {
gp[[i]] <- 10^gl[[i]] / sum(10^gl[[i]], na.rm=TRUE)
}
gp
} else if (is.array(gl) & length(dim(gl)) == 3) {
aperm(apply(gl, c(1,2), function(x)
10^x / sum(10^x, na.rm=TRUE)),
c(2,3,1))
} else {
stop("gl must be a matrix of lists or a 3D array")
}
}
.matrixOfListsToArray <- function(x) {
# find number of elements of each cell of x
n <- elementLengths(x)
maxn <- max(n)
# for cells with less than the max number of elements, add NAs
idx <- n < maxn
x[idx] <- lapply(x[idx], function(a){c(a, rep(NA, maxn-length(a)))})
# unlist and convert to array
x <- array(unlist(x), dim=c(maxn, nrow(x), ncol(x)),
dimnames=list(NULL, rownames(x), colnames(x)))
x <- aperm(x, c(2,3,1))
x
}
|
