1. celda
  2. Commit 97a92489ea2f3dc425b17bd4f849db3e0925c5de
Browse code

add phenotype enrichment plot #22

zhewa authored on 17/02/2019 21:25:14
Showing 4 changed files
DESCRIPTION
History View file @ 97a92489
... ... 
@@ -33,13 +33,14 @@ Imports:
33 33 
     Rcpp,
34 34 
     RcppEigen,
35 35 
     umap,
36 
-    enrichR
36 
+    enrichR,
37 
+    vcd,
38 
+    ggpubr
37 39 
 Suggests:
38 40 
     testthat,
39 41 
     knitr,
40 42 
     roxygen2,
41 43 
     rmarkdown,
42 
-    vcd,
43 44 
     corrplot,
44 45 
     Matrix,
45 46 
     biomaRt,
NAMESPACE
History View file @ 97a92489
... ... 
@@ -42,6 +42,7 @@ export(plotGridSearchPerplexity.celda_C)
42 42 
 export(plotGridSearchPerplexity.celda_CG)
43 43 
 export(plotGridSearchPerplexity.celda_G)
44 44 
 export(plotHeatmap)
45 
+export(plotPhenoEnrich)
45 46 
 export(recodeClusterY)
46 47 
 export(recodeClusterZ)
47 48 
 export(resList)
R/phenoEnrich.R
History View file @ 97a92489
48 49 
new file mode 100644
... ... 
@@ -0,0 +1,133 @@
1 
+#' @title Phenotype enrichment plot
2 
+#' @description Plot the enrichment of phenotypes within cell clusters or
3 
+#'  \emph{vice versa} in a given celda model. Generates a mosaic plot or
4 
+#'  a balloon plot.
5 
+#' @param counts Integer matrix. Rows represent features and columns represent
6 
+#'  cells.
7 
+#' @param celda.mod Celda model returned by the \link{celda_C},
8 
+#'  \code{\link{celda_CG}} or \code{\link{subsetCeldaList}} functions. Must be
9 
+#'  a \linkS4class{celda_C} or \linkS4class{celda_CG} object.
10 
+#' @param phenoLabels Character vector of phenotype labels for the cells. Must
11 
+#'  have the same length as the column number of count matrix \code{counts}.
12 
+#' @param phenotype The phenotypes to inspect. If specified, cells will be
13 
+#'  categorized into two groups based on whether their phenotype is in this
14 
+#'  phenotype group or not. By default, no phenotype grouping is performed.
15 
+#' @param cellCluster The cell clusters to inspect. If specified, cells will be
16 
+#'  categorized into two groups based on whether their assigned cell cluster is
17 
+#'  in this cell cluster group or not. By default, no cell cluster grouping is
18 
+#'  performed.
19 
+#' @param perspective The comparing perspective. One of "phenotype" or
20 
+#'  "cellcluster". Indicates whether to look at the enrichment of phenotypes in
21 
+#'  cell clusters or the enrichment of cell clusters in phenotypes. Default is
22 
+#'  "phenotype", which looks at the enrichment of phenotype in cell clusters.
23 
+#' @param plot The plot of visualizing the result. One of "mosaic" or "balloon".
24 
+#'  Uses the \code{\link[vcd]{mosaic}} function for plotting mosaic plot and the
25 
+#'  \code{\link[ggpubr]{ggballoonplot}} function for plotting the balloon plot.
26 
+#' @param ... Additional parameters passed to the \code{\link[vcd]{mosaic}} or
27 
+#'  \code{\link[ggpubr]{ggballoonplot}} function.
28 
+#' @return NULL if \code{plot = "mosaic"} and a ggplot object if
29 
+#'  \code{plot = "balloon"}
30 
+#' @examples
31 
+#' # Use sample labels as a surrogate for phenotype labels
32 
+#' plotPhenoEnrich(celda.CG.sim$counts,
33 
+#'     celda.CG.mod,
34 
+#'     celda.CG.sim$sample.label)
35 
+#'
36 
+#' # Look at the enrichment of "Sample_1" in cell cluster 1 in balloon plot
37 
+#' plotPhenoEnrich(counts = celda.CG.sim$counts,
38 
+#'     celda.mod = celda.CG.mod,
39 
+#'     phenoLabels = celda.CG.sim$sample.label,
40 
+#'     phenotype = "Sample_1",
41 
+#'     cellCluster = 1,
42 
+#'     plot = "balloon")
43 
+#'
44 
+#' # Look at the enrichment of cell cluster 1 & 2 in
45 
+#' # phenotype "Sample_3" & "Sample_4" in mosaic plot
46 
+#' plotPhenoEnrich(counts = celda.CG.sim$counts,
47 
+#'     celda.mod = celda.CG.mod,
48 
+#'     phenoLabels = celda.CG.sim$sample.label,
49 
+#'     phenotype = c("Sample_3", "Sample_4"),
50 
+#'     cellCluster = c(1, 2),
51 
+#'     perspective = "cellcluster")
52 
+#' @export
53 
+plotPhenoEnrich <- function(counts,
54 
+    celda.mod,
55 
+    phenoLabels,
56 
+    phenotype = NULL,
57 
+    cellCluster = NULL,
58 
+    perspective = c("phenotype", "cellcluster"),
59 
+    plot = c("mosaic", "balloon"),
60 
+    ...) {
61 
+
62 
+    if (!(is(celdaModel, "celda_CG") || is(celdaModel, "celda_C"))) {
63 
+        stop("celda.mod must be of class 'celda_C' or 'celda_CG'")
64 
+    }
65 
+
66 
+    if (length(phenoLabels) != ncol(counts)) {
67 
+        stop("phenoLabels must be the same length as the column number of",
68 
+            " counts")
69 
+    }
70 
+
71 
+    if (!is.null(phenotype)) {
72 
+        if (!(all(phenotype %in% phenoLabels))) {
73 
+            stop("phenotype ",
74 
+                paste(phenotype, collapse = " "),
75 
+                " must exist in phenoLabels")
76 
+        }
77 
+    }
78 
+
79 
+    if (!is.null(cellCluster)) {
80 
+        if (!(all(cellCluster %in% unique(celda::clusters(celda.mod)$z)))) {
81 
+            stop("Cell cluster ",
82 
+                paste(cellCluster, collapse = " "),
83 
+                " must exist in celda.mod@clusters$z")
84 
+        }
85 
+    }
86 
+
87 
+    plot <- match.arg(plot)
88 
+    perspective <- match.arg(perspective)
89 
+
90 
+    phenoLabels <- factor(phenoLabels, levels = unique(phenoLabels))
91 
+
92 
+    if (is.null(phenotype) && is.null(cellCluster)) {
93 
+        conti <- table(phenoLabels,
94 
+            celda::clusters(celda.mod)$z,
95 
+            dnn = c("Phenotype", "Cell cluster"))
96 
+    } else if (!is.null(phenotype) && is.null(cellCluster)) {
97 
+        conti <- table(phenoLabels %in% phenotype,
98 
+            celda::clusters(celda.mod)$z,
99 
+            dnn = c(paste0("In phenotype ", paste(phenotype, collapse = " ")),
100 
+                "Cell cluster"))
101 
+    } else if (is.null(phenotype) && !is.null(cellCluster)) {
102 
+        conti <- table(phenoLabels,
103 
+            celda::clusters(celda.mod)$z %in% cellCluster,
104 
+            dnn = c("Phenotype",
105 
+                paste0("In cell cluster ",
106 
+                    paste(cellCluster,
107 
+                        collapse = " "))))
108 
+    } else {
109 
+        conti <- table(phenoLabels %in% phenotype,
110 
+            celda::clusters(celda.mod)$z %in% cellCluster,
111 
+            dnn = c(paste0("In phenotype ", paste(phenotype, collapse = " ")),
112 
+                paste0("In cell cluster ", paste(cellCluster,
113 
+                    collapse = " "))))
114 
+    }
115 
+
116 
+    if (perspective == "cellcluster") {
117 
+        conti <- t(conti)
118 
+    }
119 
+
120 
+    if (plot == "mosaic") {
121 
+        vcd::mosaic(t(conti),
122 
+            shade = TRUE,
123 
+            legend = TRUE,
124 
+            direction = "v",
125 
+            ...)
126 
+    } else if (plot == "balloon") {
127 
+        g <- ggpubr::ggballoonplot(as.data.frame(t(conti)),
128 
+            fill = "value",
129 
+            ...) +
130 
+            ggplot2::scale_fill_viridis_c(option = "C")
131 
+        return(g)
132 
+    }
133 
+}
man/plotPhenoEnrich.Rd
History View file @ 97a92489
0 134 
new file mode 100644
... ... 
@@ -0,0 +1,74 @@
1 
+% Generated by roxygen2: do not edit by hand
2 
+% Please edit documentation in R/phenoEnrich.R
3 
+\name{plotPhenoEnrich}
4 
+\alias{plotPhenoEnrich}
5 
+\title{Phenotype enrichment plot}
6 
+\usage{
7 
+plotPhenoEnrich(counts, celda.mod, phenoLabels, phenotype = NULL,
8 
+  cellCluster = NULL, perspective = c("phenotype", "cellcluster"),
9 
+  plot = c("mosaic", "balloon"), ...)
10 
+}
11 
+\arguments{
12 
+\item{counts}{Integer matrix. Rows represent features and columns represent
13 
+cells.}
14 
+
15 
+\item{celda.mod}{Celda model returned by the \link{celda_C},
16 
+\code{\link{celda_CG}} or \code{\link{subsetCeldaList}} functions. Must be
17 
+a \linkS4class{celda_C} or \linkS4class{celda_CG} object.}
18 
+
19 
+\item{phenoLabels}{Character vector of phenotype labels for the cells. Must
20 
+have the same length as the column number of count matrix \code{counts}.}
21 
+
22 
+\item{phenotype}{The phenotypes to inspect. If specified, cells will be
23 
+categorized into two groups based on whether their phenotype is in this
24 
+phenotype group or not. By default, no phenotype grouping is performed.}
25 
+
26 
+\item{cellCluster}{The cell clusters to inspect. If specified, cells will be
27 
+categorized into two groups based on whether their assigned cell cluster is
28 
+in this cell cluster group or not. By default, no cell cluster grouping is
29 
+performed.}
30 
+
31 
+\item{perspective}{The comparing perspective. One of "phenotype" or
32 
+"cellcluster". Indicates whether to look at the enrichment of phenotypes in
33 
+cell clusters or the enrichment of cell clusters in phenotypes. Default is
34 
+"phenotype", which looks at the enrichment of phenotype in cell clusters.}
35 
+
36 
+\item{plot}{The plot of visualizing the result. One of "mosaic" or "balloon".
37 
+Uses the \code{\link[vcd]{mosaic}} function for plotting mosaic plot and the
38 
+\code{\link[ggpubr]{ggballoonplot}} function for plotting the balloon plot.}
39 
+
40 
+\item{...}{Additional parameters passed to the \code{\link[vcd]{mosaic}} or
41 
+\code{\link[ggpubr]{ggballoonplot}} function.}
42 
+}
43 
+\value{
44 
+NULL if \code{plot = "mosaic"} and a ggplot object if
45 
+ \code{plot = "balloon"}
46 
+}
47 
+\description{
48 
+Plot the enrichment of phenotypes within cell clusters or
49 
+ \emph{vice versa} in a given celda model. Generates a mosaic plot or
50 
+ a balloon plot.
51 
+}
52 
+\examples{
53 
+# Use sample labels as a surrogate for phenotype labels
54 
+plotPhenoEnrich(celda.CG.sim$counts,
55 
+    celda.CG.mod,
56 
+    celda.CG.sim$sample.label)
57 
+
58 
+# Look at the enrichment of "Sample_1" in cell cluster 1 in balloon plot
59 
+plotPhenoEnrich(counts = celda.CG.sim$counts,
60 
+    celda.mod = celda.CG.mod,
61 
+    phenoLabels = celda.CG.sim$sample.label,
62 
+    phenotype = "Sample_1",
63 
+    cellCluster = 1,
64 
+    plot = "balloon")
65 
+
66 
+# Look at the enrichment of cell cluster 1 & 2 in
67 
+# phenotype "Sample_3" & "Sample_4" in mosaic plot
68 
+plotPhenoEnrich(counts = celda.CG.sim$counts,
69 
+    celda.mod = celda.CG.mod,
70 
+    phenoLabels = celda.CG.sim$sample.label,
71 
+    phenotype = c("Sample_3", "Sample_4"),
72 
+    cellCluster = c(1, 2),
73 
+    perspective = "cellcluster")
74 
+}