Merge pull request #46 from YosefLab/staging

Staging

DESCRIPTION

@@ -1,6 +1,6 @@
 Package: VISION
 Title: Functional interpretation of single cell RNA-seq latent manifolds
-Version: 1.0.0
+Version: 1.0.1
 Authors@R: c(person("Matt", "Jones", email = "[email protected]", role = c("aut", "cre")),
 			 person("David", "Detomaso", email = "[email protected]", role = c("aut", "cre")),
 			 person("Tal", "Ashuach", email = "[email protected]", role = c("aut")),

NEWS.md

@@ -1,3 +1,9 @@
+# VISION 1.0.1
+
+* Bugfixes related to caching in the output report
+* Change default `cellsPerPartition` value to 10
+* Change default `filterThreshold` parameter value to 5% in applyMicroClustering (for consistency with Vision constructor)
+
 # VISION 1.0.0
 
 * Added Layout Options to the output report

R/AnalysisFunctions.R

@@ -121,16 +121,41 @@ filterData <- function(object,
     object@projection_genes <- projection_genes
     object@threshold <- threshold
 
+    message("Determining projection genes...")
+
     if (length(object@projection_genes) == 1){
-        message("Determining projection genes...")
 
         exprData <- matLog2(object@exprData)
-        object@projection_genes <- applyFilters(
+        projection_genes <- applyFilters(
                     exprData,
                     object@threshold,
                     object@projection_genes)
+
+        if (length(projection_genes) == 0){
+            stop(
+                sprintf("Filtering with (projection_genes=\"%s\", threshold=%i) results in 0 genes\n  Set a lower threshold and re-run",
+                    object@projection_genes, object@threshold)
+                )
+        }
+    } else {
+        projection_genes <- intersect(
+            object@projection_genes, rownames(object@exprData))
+
+        if (length(projection_genes) == 0){
+            stop("Supplied list of genes in `projection_genes` does not match any rows of expression data")
+        } else {
+            message(
+                sprintf("    Using supplied list of genes: Found %i/%i matches",
+                    length(projection_genes), length(object@projection_genes)
+                    )
+                )
+        }
     }
 
+    message()
+
+    object@projection_genes <- projection_genes
+
 
     return(object)
 }
@@ -228,7 +253,7 @@ computeLatentSpace <- function(object, projection_genes = NULL,
     perm_wPCA <- object@perm_wPCA
 
     if (!is.null(projection_genes)) {
-        exprData <- expr[projection_genes, ]
+        exprData <- expr[projection_genes, , drop = FALSE]
     } else {
         exprData <- expr
     }

R/Filters.R

@@ -50,13 +50,13 @@ filterGenesNovar <- function(data) {
 #' @return character vector of gene names passing filter
 filterGenesThreshold <- function(data, threshold) {
     message(
-        sprintf("    Applying Threshold filter...removing genes detected in less than %i cells", 2343)
+        sprintf("    Applying Threshold filter...removing genes detected in less than %i cells", threshold)
     )
 
     if ( is(data, "sparseMatrix") ){
-        valid_rows <- rowSums(data > 0) > threshold
+        valid_rows <- rowSums(data > 0) >= threshold
     } else {
-        valid_rows <- matrixStats::rowCounts(data > 0) > threshold
+        valid_rows <- matrixStats::rowCounts(data > 0) >= threshold
     }
 
     genes_passing <- rownames(data)[valid_rows]

R/Microclusters.R

@@ -11,10 +11,10 @@
 #'
 #'
 #' @param exprData the expression data matrix
-#' @param cellsPerPartition control over the minimum number of cells to put into each supercell
+#' @param cellsPerPartition control over the target number of cells to put into each supercell
 #' @param filterInput name of filtering method ('threshold' or 'fano') or list of
 #' genes to use when computing projections.
-#' @param filterThreshold Threshold to apply when using the 'threshold' projection genes filter.
+#' @param filterThreshold Threshold to apply when using the 'threshold' or 'fano' projection genes filter.
 #' If greater than 1, this specifies the number of cells in which a gene must be detected
 #' for it to be used when computing PCA. If less than 1, this instead specifies the proportion of cells needed
 #' @param latentSpace (Optional) Latent space to be used instead of PCA numeric matrix cells x components
@@ -23,27 +23,46 @@
 #' @return pooled cells - named list of vectors - cells in each supercell
 #' @export
 applyMicroClustering <- function(
-                         exprData, cellsPerPartition=100,
+                         exprData, cellsPerPartition=10,
                          filterInput = "fano",
-                         filterThreshold = round(ncol(exprData) * 0.2),
+                         filterThreshold = round(ncol(exprData) * 0.05),
                          latentSpace = NULL) {
 
+    if (is.data.frame(exprData)){
+        exprData <- data.matrix(exprData)
+    }
+
     if (is.null(latentSpace) || all(dim(latentSpace) == c(1, 1))) {
         exprData <- matLog2(exprData)
 
 
-        if (length(filterInput > 1)){
-            gene_passes <- filterInput
+        message("    Computing a latent space for microclustering using PCA...")
+        if (length(filterInput) > 1){
+            gene_passes <- intersect(filterInput, rownames(exprData))
+            if (length(gene_passes) == 0){
+                stop("Supplied list of genes in `filterInput` does not match any rows of `exprData`")
+            } else {
+                message(
+                    sprintf("    Using supplied list of genes: Found %i/%i matches", length(gene_passes), length(filterInput))
+                    )
+            }
         } else {
+            message("    Determining lateng space genes...")
             gene_passes <- applyFilters(exprData, filterThreshold, filterInput)
+
+            if (length(gene_passes) == 0){
+                stop(
+                    sprintf("Filtering with (filterInput=\"%s\", filterThreshold=%i) results in 0 genes\n  Set a lower threshold and re-run", filterInput, filterThreshold)
+                    )
+            }
         }
 
-        fexpr <- exprData[gene_passes, ]
+        fexpr <- exprData[gene_passes, , drop = FALSE]
 
-        message("    Computing a latent space for microclustering using PCA...")
         # Compute wcov using matrix operations to avoid
         # creating a large dense matrix
 
+        message("    Performing PCA...")
         N <- ncol(fexpr)
         wcov <- tcrossprod(fexpr) / N
 
@@ -82,7 +101,7 @@ applyMicroClustering <- function(
     pools <- readjust_clusters(cl, res, cellsPerPartition = cellsPerPartition)
 
     # Rename clusters
-    cn <- paste0("microcluster ", 1:length(pools))
+    cn <- paste0("microcluster_", 1:length(pools))
     names(pools) <- cn
 
     message(

R/Projections.R

@@ -45,7 +45,7 @@ generateProjectionsInner <- function(expr, latentSpace, projection_genes=NULL, p
         if (method == "ICA" || method == "RBFPCA") {
 
             if (!is.null(projection_genes)) {
-                exprData <- expr[projection_genes, ]
+                exprData <- expr[projection_genes, , drop = FALSE]
             } else {
                 exprData <- expr
             }

R/Server.R

@@ -31,7 +31,7 @@ ServerSigProjMatrix <- function(zscores, pvals, proj_labels, sig_labels) {
 #' @return JSON formatted Signature object.
 signatureToJSON <- function(sig) {
 
-    # Pass in a Signature object from a FastProject Object to be converted into a JSON object
+    # Pass in a Signature object from an R Object to be converted into a JSON object
     sig@sigDict <- as.list(sig@sigDict)
 
     json <- toJSON(sig, force=TRUE, pretty=TRUE, auto_unbox=TRUE)
@@ -100,7 +100,7 @@ coordinatesToJSON <- function(p) {
     return(json)
 }
 
-#' Converts a sigProjMatrix from a FastProject Object to a JSON object
+#' Converts a sigProjMatrix from an R Object to a JSON object
 #' @importFrom jsonlite toJSON
 #' @param sigzscores Matrix of signature z-scores
 #' @param sigpvals Matrix of signature p-values
@@ -156,8 +156,7 @@ compressJSONResponse <- function(json, res, req){
 #' Lanch the server
 #' @importFrom jsonlite fromJSON
 #' @importFrom utils browseURL URLdecode stack
-#' @param object FastProject object or path to a file containing such an
-#' object (saved using saveAndViewResults, or directly using saveRDS)
+#' @param object Vision object
 #' @param port The port on which to serve the output viewer.  If omitted, a
 #' random port between 8000 and 9999 is chosen.
 #' @param host The host used to serve the output viewer. If omitted, "127.0.0.1"

R/SigScoreMethods.R

@@ -1,10 +1,10 @@
 #' Different ways to evalutate the signature score
 #'
-#' EAch method should have the same signature so they can be swapped
+#' Each method should have the same signature so they can be swapped
 #'
 #' Right now, each takes in a wrapped data object and signature object
 #'
-#' Specified by FastProject argument (sig_score_method), default = naiveEvalSignature
+#' Specified by Vision argument (sig_score_method), default = "naive"
 
 #' Evaluate signature scores efficiently in batches
 #'

R/Utilities.R

@@ -35,15 +35,15 @@ batchify <- function(items, per_batch, n_workers = 1) {
 }
 
 
-#' Check's the version of the FastProject object and displays error if necessary
+#' Checks the version of the Vision object and displays error if necessary
 #'
-#' @param object FastProject object
+#' @param object Vision object
 #' @return NULL
 versionCheck <- function(object) {
 
     templateStr <- paste0(
-        "This FastProject object was created with an older version of the library.",
-        "  To view, either install an older version (commit #COMMIT) from GitHub (www.github.com/YosefLab/FastProjectR) or",
+        "This Vision object was created with an older version of the library.",
+        "  To view, either install an older version (commit #COMMIT) from GitHub (www.github.com/YosefLab/Vision) or",
         " recreate the object and re-run analyze"
     )
 
@@ -57,6 +57,11 @@ versionCheck <- function(object) {
         stop(msg, call. = FALSE)
     }
 
+    if(object@version < 1.1) {
+        msg <- gsub("#COMMIT", "2db4552", templateStr)
+        stop(msg, call. = FALSE)
+    }
+
     # Add new commit hashes here as version increases and breaks backward compatibility
 
     return()

R/methods-Vision.R

@@ -23,7 +23,7 @@
 #' supplied expression matrix are removed.
 #' @param weights Precomputed weights for each coordinate. Normally computed
 #' from the FNR curve.
-#' @param threshold Threshold to apply when using the 'threshold' projection genes filter.
+#' @param threshold Threshold to apply when using the 'threshold' or 'fano' projection genes filter.
 #' If greater than 1, this specifies the number of cells in which a gene must be detected
 #' for it to be used when computing PCA. If less than 1, this instead specifies the proportion of cells needed
 #' @param perm_wPCA If TRUE, apply permutation procedure to calculate significant
@@ -34,10 +34,10 @@
 #' or "rank_norm_columns"
 #' @param sig_score_method Method to apply when calculating signature scores.
 #' Either "naive" (default) or "weighted_avg"
-#' @param pool indicates whether or not to create supercells. Acceptable values
+#' @param pool indicates whether or not to pool cells into supercells. Acceptable values
 #' are TRUE, FALSE, or 'auto', the last of which is the default and enables
 #' pooling if there are more than 15000 cells.
-#' @param cellsPerPartition the minimum number of cells to put into a cluster
+#' @param cellsPerPartition the target number of cells to put into a supercell when pooling
 #' @param latentSpace latent space for expression data. Numeric matrix or dataframe
 #' with dimensions CELLS x COMPONENTS
 #' @param latentTrajectory trajectory to model cell progression.  Wrapped result
@@ -78,7 +78,7 @@ setMethod("Vision", signature(data = "matrixORSparse"),
                                         "znorm_rows_then_columns",
                                         "rank_norm_columns"),
                     sig_score_method=c("naive", "weighted_avg"),
-                    pool="auto", cellsPerPartition=100, name=NULL,
+                    pool="auto", cellsPerPartition=10, name=NULL,
                     latentSpace = NULL, latentTrajectory = NULL, pools=list()) {
 
             .Object <- new("Vision")