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
|
% Generated by roxygen2: do not edit by hand
% Please edit documentation in R/perCellQCMetrics.R
\name{perCellQCMetrics}
\alias{perCellQCMetrics}
\alias{perCellQCMetrics,ANY-method}
\alias{perCellQCMetrics,SummarizedExperiment-method}
\alias{perCellQCMetrics,SingleCellExperiment-method}
\title{Per-cell quality control metrics}
\usage{
perCellQCMetrics(x, ...)
\S4method{perCellQCMetrics}{ANY}(
x,
subsets = NULL,
percent.top = integer(0),
threshold = 0,
BPPARAM = SerialParam(),
flatten = TRUE,
percent_top = NULL,
detection_limit = NULL
)
\S4method{perCellQCMetrics}{SummarizedExperiment}(x, ..., assay.type = "counts", exprs_values = NULL)
\S4method{perCellQCMetrics}{SingleCellExperiment}(
x,
subsets = NULL,
percent.top = integer(0),
...,
flatten = TRUE,
assay.type = "counts",
use.altexps = NULL,
percent_top = NULL,
exprs_values = NULL,
use_altexps = NULL
)
}
\arguments{
\item{x}{A numeric matrix of counts with cells in columns and features in rows.
Alternatively, a \linkS4class{SummarizedExperiment} or \linkS4class{SingleCellExperiment} object containing such a matrix.}
\item{...}{For the generic, further arguments to pass to specific methods.
For the SummarizedExperiment and SingleCellExperiment methods, further arguments to pass to the ANY method.}
\item{subsets}{A named list containing one or more vectors
(a character vector of feature names, a logical vector, or a numeric vector of indices),
used to identify interesting feature subsets such as ERCC spike-in transcripts or mitochondrial genes.}
\item{percent.top}{An integer vector specifying the size(s) of the top set of high-abundance genes.
Used to compute the percentage of library size occupied by the most highly expressed genes in each cell.}
\item{threshold}{A numeric scalar specifying the threshold above which a gene is considered to be detected.}
\item{BPPARAM}{A \linkS4class{BiocParallelParam} object specifying how parallelization should be performed.}
\item{flatten}{Logical scalar indicating whether the nested \linkS4class{DataFrame}s in the output should be flattened.}
\item{percent_top, detection_limit, exprs_values, use_altexps}{Soft deprecated equivalents to the arguments described above.}
\item{assay.type}{A string or integer scalar indicating which \code{assays} in the \code{x} contains the count matrix.}
\item{use.altexps}{Logical scalar indicating whether QC statistics should be computed for alternative Experiments in \code{x}.
If \code{TRUE}, statistics are computed for all alternative experiments.
Alternatively, an integer or character vector specifying the alternative Experiments to use to compute QC statistics.
Alternatively \code{NULL}, in which case we only use alternative experiments that contain the specified \code{assay.type}.
Alternatively \code{FALSE}, in which case alternative experiments are not used.}
}
\value{
A \linkS4class{DataFrame} of QC statistics where each row corresponds to a column in \code{x}.
This contains the following fields:
\itemize{
\item \code{sum}: numeric, the sum of counts for each cell.
\item \code{detected}: numeric, the number of observations above \code{threshold}.
}
If \code{flatten=FALSE}, the DataFrame will contain the additional columns:
\itemize{
\item \code{percent.top}: numeric matrix, the percentage of counts assigned to the top most highly expressed genes.
Each column of the matrix corresponds to an entry of \code{percent.top}, sorted in increasing order.
\item \code{subsets}: A nested DataFrame containing statistics for each subset, see Details.
\item \code{altexps}: A nested DataFrame containing statistics for each alternative experiment, see Details.
This is only returned for the SingleCellExperiment method.
\item \code{total}: numeric, the total sum of counts for each cell across main and alternative Experiments.
This is only returned for the SingleCellExperiment method.
}
If \code{flatten=TRUE}, nested matrices and DataFrames are flattened to remove the hierarchical structure from the output DataFrame.
}
\description{
Compute per-cell quality control metrics for a count matrix or a \linkS4class{SingleCellExperiment}.
}
\details{
This function calculates useful QC metrics for identification and removal of potentially problematic cells.
Obvious per-cell metrics are the sum of counts (i.e., the library size) and the number of detected features.
The percentage of counts in the top features also provides a measure of library complexity.
If \code{subsets} is specified, these statistics are also computed for each subset of features.
This is useful for investigating gene sets of interest, e.g., mitochondrial genes, Y chromosome genes.
These statistics are stored as nested \linkS4class{DataFrame}s in the \code{subsets} field of the output.
For example, if the input \code{subsets} contained \code{"Mito"} and \code{"Sex"}, the output would look like:
\preformatted{ output
|-- sum
|-- detected
|-- percent.top
+-- subsets
|-- Mito
| |-- sum
| |-- detected
| +-- percent
+-- Sex
|-- sum
|-- detected
+-- percent
}
Here, the \code{percent} field contains the percentage of each cell's count \code{sum} assigned to each subset.
If \code{use.altexps=TRUE}, the same statistics are computed for each alternative experiment in \code{x}.
This can also be an integer or character vector specifying the alternative Experiments to use.
These statistics are also stored as nested \linkS4class{DataFrame}s, this time in the \code{altexps} field of the output.
For example, if \code{x} contained the alternative Experiments \code{"Spike"} and \code{"Ab"}, the output would look like:
\preformatted{ output
|-- sum
|-- detected
|-- percent.top
+-- altexps
| |-- Spike
| | |-- sum
| | |-- detected
| | +-- percent
| +-- Ab
| |-- sum
| |-- detected
| +-- percent
+-- total
}
The \code{total} field contains the total sum of counts for each cell across the main and alternative Experiments.
The \code{percent} field contains the percentage of the \code{total} count in each alternative Experiment for each cell.
Note that the denominator for \code{altexps$...$percent} is not the same as the denominator for \code{subset$...$percent}.
For example, if \code{subsets} contains a set of mitochondrial genes, the mitochondrial percentage would be computed as a fraction of the total endogenous coverage,
while the \code{altexps} percentage would be computed as a fraction of the total coverage across all (endogenous and artificial) features.
If \code{flatten=TRUE}, the nested DataFrames are flattened by concatenating the column names with underscores.
This means that, say, the \code{subsets$Mito$sum} nested field becomes the top-level \code{subsets_Mito_sum} field.
A flattened structure is more convenient for end-users performing interactive analyses,
but less convenient for programmatic access as artificial construction of strings is required.
}
\examples{
example_sce <- mockSCE()
stats <- perCellQCMetrics(example_sce)
stats
# With subsets.
stats2 <- perCellQCMetrics(example_sce, subsets=list(Mito=1:10),
flatten=FALSE)
stats2$subsets
# With alternative Experiments.
pretend.spike <- ifelse(seq_len(nrow(example_sce)) < 10, "Spike", "Gene")
alt_sce <- splitAltExps(example_sce, pretend.spike)
stats3 <- perCellQCMetrics(alt_sce, flatten=FALSE)
stats3$altexps
}
\seealso{
\code{\link{addPerCellQCMetrics}}, to add the QC metrics to the column metadata.
}
\author{
Aaron Lun
}
|
