Merge pull request #301 from mvfki/newObj

fixes

R/DoubletFinder.R

@@ -27,8 +27,10 @@
 #' \code{DoubletFinder_classification} updated in \code{cellMeta} slot
 #' @export
 #' @examples
-#' pbmc <- runDoubletFinder(pbmc)
-#' cellMeta(pbmc)
+#' if (requireNamespace("DoubletFinder", quietly = TRUE)) {
+#'     pbmc <- runDoubletFinder(pbmc)
+#'     print(cellMeta(pbmc))
+#' }
 runDoubletFinder <- function(
         object,
         useDatasets = NULL,

R/GSEA.R

@@ -16,7 +16,9 @@
 #' @return A list of matrices with GSEA analysis for each factor
 #' @export
 #' @examples
+#' \donttest{
 #' runGSEA(pbmcPlot)
+#' }
 runGSEA <- function(
         object,
         genesets = NULL,

R/clustering.R

@@ -1,13 +1,14 @@
 #' SNN Graph Based Community Detection
-#' @description After quantile normalization, users can additionally run the
-#' Leiden or Louvain algorithm for community detection, which is widely used in
+#' @description
+#' After quantile normalization, users can additionally run the Leiden or
+#' Louvain algorithm for community detection, which is widely used in
 #' single-cell analysis and excels at merging small clusters into broad cell
 #' classes.
 #'
 #' While using quantile normalized factor loadings (result from
 #' \code{\link{quantileNorm}}) is recommended, this function looks for
-#' unnormalized factor loadings (result from \code{\link{optimizeALS}} or
-#' \code{\link{online_iNMF}}) when the former is not available.
+#' unnormalized factor loadings (result from \code{\link{runIntegration}}) when
+#' the former is not available.
 #' @param object A \linkS4class{liger} object. Should have valid factorization
 #' result available.
 #' @param nNeighbors Integer, the maximum number of nearest neighbors to
@@ -28,6 +29,9 @@
 #' Default \code{5}.
 #' @param method Community detection algorithm to use. Choose from
 #' \code{"leiden"} or \code{"louvain"}. Default \code{"leiden"}.
+#' @param useRaw Whether to use un-aligned cell factor loadings (\eqn{H}
+#' matrices). Default \code{NULL} search for quantile-normalized loadings first
+#' and un-aligned loadings then.
 #' @param useDims Indices of factors to use for clustering. Default \code{NULL}
 #' uses all available factors.
 #' @param groupSingletons Whether to group single cells that make up their own
@@ -58,6 +62,7 @@ runCluster <- function(
         nRandomStarts = 10,
         nIterations = 5,
         method = c("leiden", "louvain"),
+        useRaw = NULL,
         useDims = NULL,
         groupSingletons = TRUE,
         clusterName = paste0(method, "_cluster"),
@@ -69,21 +74,16 @@ runCluster <- function(
         leiden = recordCommand(object, dependencies = c("RANN", "leidenAlg")),
         louvain = recordCommand(object, dependencies = c("RANN"))
     )
+    Hsearch <- searchH(object, useRaw)
+    H <- Hsearch$H
+    useRaw <- Hsearch$useRaw
+    type <- ifelse(useRaw, " unnormalized ", " quantile normalized ")
 
-    H.norm <- getMatrix(object, "H.norm")
-    if (is.null(H.norm)) {
-        type <- " unnormalized "
-        H.norm <- Reduce(cbind, getMatrix(object, "H"))
-    } else type <- " quantile normalized "
-    if (is.null(H.norm))
-        stop("No factor loading ('H.norm' or 'H') found in `object`.")
-    if (type == " unnormalized ") H.norm <- t(H.norm)
-
-    if (!is.null(useDims)) H.norm <- H.norm[, useDims]
+    if (!is.null(useDims)) H <- H[, useDims, drop = FALSE]
 
     if (isTRUE(verbose))
         .log(method, " clustering on", type, "cell factor loadings...")
-    knn <- RANN::nn2(H.norm, k = nNeighbors, eps = eps)
+    knn <- RANN::nn2(H, k = nNeighbors, eps = eps)
     snn <- ComputeSNN(knn$nn.idx, prune = prune)
     if (!is.null(seed)) set.seed(seed)
     if (method == "leiden") {
@@ -93,7 +93,7 @@ runCluster <- function(
         edge_weights <- snnSummary[,3]
         clusts <- leidenAlg::find_partition_with_rep_rcpp(
             edgelist = edgelist, edgelist_length = edgelist_length,
-            num_vertices = nrow(H.norm), direction = FALSE,
+            num_vertices = nrow(H), direction = FALSE,
             edge_weights = edge_weights, resolution = resolution,
             niter = nIterations, nrep = nRandomStarts
         )

R/data.R

@@ -7,10 +7,10 @@
 
 #' liger object of PBMC subsample data with plotting information available
 #' @description This data was generated from data \code{"pbmc"} with default
-#' parameter integration pipeline, with setting \code{k = 20, maxIter = 10}
-#' for \code{\link{optimizeALS}}. UMAP was generated with \code{minDist = 0.5}.
-#' Only the 2nd and 3rd factors, the normalized expression of the top 50
-#' variable genes, and the clustering label were kept for visualization example.
+#' parameter integration pipeline: normalize, selectGenes, scaleNotCenter,
+#' runINMF, runCluster, runUMAP. To minimize the object size distributed with
+#' the package, rawData and scaleData were removed. Genes are downsampled to
+#' the top 50 variable genes, for smaller normData and \eqn{W} matrix.
 #' @format \linkS4class{liger} object with two datasets named by "ctrl" and
 #' "stim".
 #' @source https://www.nature.com/articles/nbt.4042

R/embedding.R

@@ -1,12 +1,12 @@
 #' Perform UMAP Dimensionality Reduction
-#' @description Run UMAP on the quantile normalized cell factors (result from
-#' \code{\link{quantileNorm}}), or unnormalized cell
-#' factors (result from \code{\link{optimizeALS}} or \code{\link{online_iNMF}}))
-#' to generate a 2D embedding for visualization (or general dimensionality
-#' reduction). Has option to run on subset of factors. It is generally
-#' recommended to use this method for dimensionality reduction with extremely
-#' large datasets. The underlying UMAP calculation imports uwot
-#' \code{\link[uwot]{umap}}.
+#' @description
+#' Run UMAP on the quantile normalized cell factors (result from
+#' \code{\link{quantileNorm}}), or unnormalized cell factors (result from
+#' \code{\link{runIntegration}})) to generate a 2D embedding for visualization
+#' (or general dimensionality reduction). Has option to run on subset of
+#' factors. It is generally recommended to use this method for dimensionality
+#' reduction with extremely large datasets. The underlying UMAP calculation
+#' imports uwot \code{\link[uwot]{umap}}.
 #' @details For \code{nNeighbors}, larger values will result in more global
 #' structure being preserved at the loss of detailed local structure. In general
 #' this parameter should often be in the range 5 to 50, with a choice of 10 to
@@ -18,7 +18,8 @@
 #' 0.001 to 0.5, with 0.1 being a reasonable default.
 #' @param object \linkS4class{liger} object with factorization results.
 #' @param useRaw Whether to use un-aligned cell factor loadings (\eqn{H}
-#' matrices). Default \code{FALSE}.
+#' matrices). Default \code{NULL} search for quantile-normalized loadings first
+#' and un-aligned loadings then.
 #' @param useDims Index of factors to use for computing UMAP embedding. Default
 #' \code{NULL} uses all factors.
 #' @param nDims Number of dimensions to reduce to. Default \code{2}.
@@ -44,7 +45,7 @@
 #' pbmc <- runUMAP(pbmcPlot)
 runUMAP <- function(
         object,
-        useRaw = FALSE,
+        useRaw = NULL,
         useDims = NULL,
         nDims = 2,
         distance = c("cosine", "euclidean", "manhattan", "hamming"),
@@ -67,16 +68,13 @@ runUMAP <- function(
     distance <- match.arg(distance)
     object <- recordCommand(object, dependencies = "uwot")
     set.seed(seed)
-    if (isTRUE(useRaw)) {
-        type <- " unnormalized "
-        H <- t(Reduce(cbind, getMatrix(object, "H")))
-    } else {
-        type <- " quantile normalized "
-        H <- getMatrix(object, "H.norm")
-    }
+    Hsearch <- searchH(object, useRaw)
+    H <- Hsearch$H
+    useRaw <- Hsearch$useRaw
+    type <- ifelse(useRaw, " unnormalized ", " quantile normalized ")
     if (isTRUE(verbose))
         .log("Generating UMAP on", type, "cell factor loadings...")
-    if (!is.null(useDims)) H <- H[, useDims]
+    if (!is.null(useDims)) H <- H[, useDims, drop = FALSE]
     umap <- uwot::umap(H,
                        n_components = as.integer(nDims),
                        metric = distance,
@@ -87,13 +85,14 @@ runUMAP <- function(
 }
 
 #' Perform t-SNE dimensionality reduction
-#' @description Runs t-SNE on the quantile normalized cell factors (result from
+#' @description
+#' Runs t-SNE on the quantile normalized cell factors (result from
 #' \code{\link{quantileNorm}}), or unnormalized cell factors (result from
-#' \code{\link{optimizeALS}} or \code{\link{online_iNMF}})) to generate a 2D
-#' embedding for visualization. By default \code{\link[Rtsne]{Rtsne}}
-#' (Barnes-Hut implementation of t-SNE) method is invoked, while alternative
-#' "fftRtsne" method (FFT-accelerated Interpolation-based t-SNE, using Kluger
-#' Lab implementation) is also supported.
+#' \code{\link{runIntegration}})) to generate a 2D embedding for visualization.
+#' By default \code{\link[Rtsne]{Rtsne}} (Barnes-Hut implementation of t-SNE)
+#' method is invoked, while alternative "fftRtsne" method (FFT-accelerated
+#' Interpolation-based t-SNE, using Kluger Lab implementation) is also
+#' supported.
 #'
 #' In order to run fftRtsne (recommended for large datasets), FIt-SNE must be
 #' installed as instructed in detailed
@@ -104,7 +103,8 @@ runUMAP <- function(
 #' \href{https://github.com/welch-lab/liger#readme}{README}.
 #' @param object \linkS4class{liger} object with factorization results.
 #' @param useRaw Whether to use un-aligned cell factor loadings (\eqn{H}
-#' matrices). Default \code{FALSE}.
+#' matrices). Default \code{NULL} search for quantile-normalized loadings first
+#' and un-aligned loadings then.
 #' @param useDims Index of factors to use for computing UMAP embedding. Default
 #' \code{NULL} uses all factors.
 #' @param nDims Number of dimensions to reduce to. Default \code{2}.
@@ -134,7 +134,7 @@ runUMAP <- function(
 #' pbmc <- runTSNE(pbmcPlot)
 runTSNE <- function(
         object,
-        useRaw = FALSE,
+        useRaw = NULL,
         useDims = NULL,
         nDims = 2,
         usePCA = FALSE,
@@ -158,28 +158,25 @@ runTSNE <- function(
                         rand.seed = "seed"))
     method <- match.arg(method)
     object <- recordCommand(object, dependencies = "Rtsne")
-    if (isTRUE(useRaw)) {
-        type <- " unnormalized "
-        data.use <- t(Reduce(cbind, getMatrix(object, "H")))
-    } else {
-        type <- " quantile normalized "
-        data.use <- getMatrix(object, "H.norm")
-    }
+    Hsearch <- searchH(object, useRaw)
+    H <- Hsearch$H
+    useRaw <- Hsearch$useRaw
+    type <- ifelse(useRaw, " unnormalized ", " quantile normalized ")
     if (isTRUE(verbose))
         .log("Generating TSNE (", method, ") on", type,
              "cell factor loadings...")
-    if (!is.null(useDims)) data.use <- data.use[, useDims]
+    if (!is.null(useDims)) H <- H[, useDims, drop = FALSE]
     if (method == "Rtsne") {
         set.seed(seed)
-        tsne <- Rtsne::Rtsne(data.use,
+        tsne <- Rtsne::Rtsne(H,
                              dims = nDims,
                              pca = usePCA,
                              check_duplicates = FALSE,
                              theta = theta,
                              perplexity = perplexity)
         tsne <- tsne$Y
     } else if (method == "fftRtsne") {
-        tsne <- .fftRtsne(data.use,
+        tsne <- .fftRtsne(H,
                           dims = nDims,
                           rand_seed = seed,
                           fast_tsne_path = fitsnePath,

R/util.R

@@ -545,3 +545,42 @@ splitRmMiss <- function(x, y) {
     names(matList) <- levels(y)
     return(matList)
 }
+
+searchH <- function(object, useRaw = NULL) {
+    if (is.null(useRaw)) {
+        # By default, look for quantile-normed H
+        H <- getMatrix(object, "H.norm")
+        if (is.null(H)) {
+            # If not found, look for raw H
+            Ht <- Reduce(cbind, getMatrix(object, "H"))
+            if (is.null(Ht)) {
+                stop("No cell factor loading available. ",
+                     "Please run `runIntegration()` and `quantileNorm()` first.")
+            } else {
+                useRaw <- TRUE
+                H <- t(Ht)
+            }
+        } else {
+            useRaw <- FALSE
+        }
+    } else {
+        if (isTRUE(useRaw)) {
+            Ht <- Reduce(cbind, getMatrix(object, "H"))
+            if (is.null(Ht)) {
+                stop("Raw cell factor loading requested but not found. ",
+                     "Please run `runIntegration()`.")
+            } else {
+                H <- t(Ht)
+            }
+        } else {
+            H <- getMatrix(object, "H.norm")
+            if (is.null(H)) {
+                stop("Quantile-normalized cell factor loading requested but ",
+                     "not found. Please run `quantileNorm()` after ",
+                     "`runIntegration()`.")
+            }
+            useRaw <- FALSE
+        }
+    }
+    return(list(H = H, useRaw = useRaw))
+}