Merge branch 'main' into readGct

ctlab · Mar 27, 2024 · 8eaa138 · 8eaa138
2 parents e33f88a + 9eeb7d8
commit 8eaa138
Show file tree

Hide file tree

Showing 13 changed files with 108 additions and 95 deletions.
diff --git a/DESCRIPTION b/DESCRIPTION
@@ -1,18 +1,18 @@
 Package: phantasusLite
 Type: Package
-Title: Loading and annotation RNA-Seq counts matrices 
-Version: 1.1.0
+Title: Loading and annotation RNA-seq counts matrices 
+Version: 1.1.1
 Authors@R: c(person("Rita", "Sablina", role = "aut"),        
              person("Maxim", "Kleverov", role = "aut"),        
              person("Alexey", "Sergushichev", email = "[email protected]", role = c("aut", "cre")))
 Description: PhantasusLite – a lightweight package with helper functions of general interest 
     extracted from phantasus package. In parituclar it simplifies working with public
     RNA-seq datasets from GEO by providing access to the remote
     HSDS repository with the precomputed gene counts from ARCHS4 and DEE2 projects.
-Depends: R (>= 4.3)
+Depends: R (>= 4.2)
 Imports: 
     data.table, 
-    rhdf5client(>= 1.21.5), 
+    rhdf5client(>= 1.25.1), 
     httr, stringr,
     stats, 
     utils, 
@@ -22,7 +22,7 @@ biocViews: GeneExpression, Transcriptomics, RNASeq
 License: MIT + file LICENSE
 Encoding: UTF-8
 LazyData: true
-RoxygenNote: 7.2.3
+RoxygenNote: 7.3.1
 Suggests: 
     testthat (>= 3.0.0), 
     knitr, 

diff --git a/NEWS b/NEWS
@@ -0,0 +1,3 @@
+Changes in version 1.2.0
+* Switch to https://alserglab.wustl.edu/hsds remote for default HSDS server
+* Depending on rhdf5client >= 1.25.1 to support ARCHS4 v2.3 files
diff --git a/R/getHSDSFileList.R b/R/getHSDSFileList.R
@@ -1,48 +1,48 @@
-#' Returns list of all HDF5-files on HSDS-server
-#' @param url, containing url of the server and root domain.
-#' @param directory, containing name of the directory
-#'
-#' @return List of all HDF5-files on the server or all files of the collection
-#'
-#' @export
-#' @import rhdf5client
-#' @examples
-#' url <- 'https://ctlab.itmo.ru/hsds/?domain=/counts'
-#' getHSDSFileList(url)
-#'
-
-getHSDSFileList <- function(url='https://ctlab.itmo.ru/hsds/?domain=/counts', directory = NULL) {
-    src <- httr::parse_url(url)
-    dir <- src$query$domain
-    src <- paste0(src$scheme,'://',src$hostname,'/',src$path)
-    src <- HSDSSource(src)
-    hdf5FileList  <- list()
-    if (is.null(directory)) {
-      directories <- listDomains(src, dir)
-      directories <- directories[-grep("*\\.h5$", directories)]
-      directories <- gsub(paste0(dir, '/'), '', directories)
-      for (directory in directories) {
-        request <- paste0(src@endpoint, "/domains?domain=",
-                          dir, '/', directory)
-        response <- rhdf5client:::submitRequest(request)
-        domains <- response[["domains"]]
-        for (domain in domains) {
-          if (domain$name != paste0(dir, "/", directory, '/', directory, ".h5")) {
-            hdf5FileList <- append(hdf5FileList, domain$name)
-          }
-        }
-      }
-    } else {
-      request <- paste0(src@endpoint, "/domains?domain=",
-                        dir, '/', directory)
-      response <- rhdf5client:::submitRequest(request)
-      domains <- response[["domains"]]
-      for (domain in domains) {
-        if (domain$name != paste0(dir, "/", directory, '/', directory, ".h5")) {
-          hdf5FileList <- append(hdf5FileList, domain$name)
-        }
-      }
-    }
-    hdf5FileList <- unlist(hdf5FileList)
-    return(hdf5FileList)
-}
+#' Returns list of all HDF5-files on HSDS-server
+#' @param url, containing url of the server and root domain.
+#' @param directory, containing name of the directory
+#'
+#' @return List of all HDF5-files on the server or all files of the collection
+#'
+#' @export
+#' @import rhdf5client
+#' @examples
+#' url <- 'https://alserglab.wustl.edu/hsds/?domain=/counts'
+#' getHSDSFileList(url)
+#'
+
+getHSDSFileList <- function(url='https://alserglab.wustl.edu/hsds/?domain=/counts', directory = NULL) {
+    src <- httr::parse_url(url)
+    dir <- src$query$domain
+    src <- paste0(src$scheme,'://',src$hostname,'/',src$path)
+    src <- HSDSSource(src)
+    hdf5FileList  <- list()
+    if (is.null(directory)) {
+      directories <- listDomains(src, dir)
+      directories <- directories[-grep("*\\.h5$", directories)]
+      directories <- gsub(paste0(dir, '/'), '', directories)
+      for (directory in directories) {
+        request <- paste0(src@endpoint, "/domains?domain=",
+                          dir, '/', directory)
+        response <- rhdf5client:::submitRequest(request)
+        domains <- response[["domains"]]
+        for (domain in domains) {
+          if (domain$name != paste0(dir, "/", directory, '/', directory, ".h5")) {
+            hdf5FileList <- append(hdf5FileList, domain$name)
+          }
+        }
+      }
+    } else {
+      request <- paste0(src@endpoint, "/domains?domain=",
+                        dir, '/', directory)
+      response <- rhdf5client:::submitRequest(request)
+      domains <- response[["domains"]]
+      for (domain in domains) {
+        if (domain$name != paste0(dir, "/", directory, '/', directory, ".h5")) {
+          hdf5FileList <- append(hdf5FileList, domain$name)
+        }
+      }
+    }
+    hdf5FileList <- unlist(hdf5FileList)
+    return(hdf5FileList)
+}
diff --git a/R/loadCountsFromH5file.R b/R/loadCountsFromH5file.R
@@ -35,10 +35,10 @@ getSamples <- function(h5f, samples_id) {
 #' @examples
 #' ess <- GEOquery::getGEO("GSE85653")
 #' es <- ess[[1]]
-#' url <- 'https://ctlab.itmo.ru/hsds/?domain=/counts'
+#' url <- 'https://alserglab.wustl.edu/hsds/?domain=/counts'
 #' file <- "/dee2/athaliana_star_matrix_20221107.h5"
 #' es <- loadCountsFromH5FileHSDS(es, url, file)
-loadCountsFromH5FileHSDS <- function(es, url='https://ctlab.itmo.ru/hsds/?domain=/counts', file, sampleIndexes = NULL) {
+loadCountsFromH5FileHSDS <- function(es, url='https://alserglab.wustl.edu/hsds/?domain=/counts', file, sampleIndexes = NULL) {
   if (nrow(es) > 0) {
     return(es)
   }
@@ -129,10 +129,10 @@ loadCountsFromH5FileHSDS <- function(es, url='https://ctlab.itmo.ru/hsds/?domain
 #' @examples
 #' ess <- GEOquery::getGEO("GSE85653")
 #' es <- ess[[1]]
-#' url <- 'https://ctlab.itmo.ru/hsds/?domain=/counts'
+#' url <- 'https://alserglab.wustl.edu/hsds/?domain=/counts'
 #' es <- loadCountsFromHSDS(es, url)
 #'
-loadCountsFromHSDS <- function(es, url='https://ctlab.itmo.ru/hsds/?domain=/counts') {
+loadCountsFromHSDS <- function(es, url='https://alserglab.wustl.edu/hsds/?domain=/counts') {
   if (nrow(es) > 0) {
     return(es)
   }

diff --git a/R/updateAndCreateMetaLocal.R b/R/updateAndCreateMetaLocal.R
@@ -35,6 +35,7 @@ createMetaH5 <- function(counts_dir){
       message("Skipping ", h5filename, " as it's already exists")
       next
     }
+    message("Creating ", h5filename)
     createH5(h5_meta, h5filename, 'meta')
   }
   return(invisible(NULL))
@@ -193,7 +194,7 @@ createIndexH5 <- function(data, file) {
   for (i in seq_along(names)) {
     rhdf5::h5write(data[[i]], file, paste0("/",names[i]))
   }
-  h5closeAll()
+  rhdf5::h5closeAll()
   return(invisible(NULL))
 }
 

diff --git a/R/updateAndCreateMetaRemote.R b/R/updateAndCreateMetaRemote.R
@@ -58,7 +58,7 @@ createIndexH5Remote <- function(url,
                                 collections=c('archs4', 'dee2'),
                                 destfile="index.h5") {
   if (file.exists(destfile)) {
-    stop("File ", destfile, " alsready exists")
+    stop("File ", destfile, " already exists")
   }
 
   DT_h5_meta <- getIndexRemote(url, collections)

diff --git a/README.md b/README.md
@@ -67,39 +67,39 @@ Function loadCountsFromHSDS returns an ExpressionSet with the expression
 matrix – the second exprs(es) contains an expression matrix.
 
 The remote repository URL is
-‘<https://ctlab.itmo.ru/hsds/?domain=/counts>’.
+‘<https://alserglab.wustl.edu/hsds/?domain=/counts>’.
 
 ``` r
-url <- 'https://ctlab.itmo.ru/hsds/?domain=/counts'
+url <- 'https://alserglab.wustl.edu/hsds/?domain=/counts'
 es <- loadCountsFromHSDS(es, url)
 head(exprs(es))
 ```
 
-    ##               GSM1281300 GSM1281301 GSM1281302 GSM1281303 GSM1281304 GSM1281305
-    ## 0610007P14Rik         86         67         30         46         23         61
-    ## 0610009B22Rik         29         22          3          0         33         13
-    ## 0610009L18Rik          0          0          7          0          0         15
-    ## 0610009O20Rik        103         38         17         20         31         54
-    ## 0610010F05Rik        259         91        115         88        113        185
-    ## 0610010K14Rik         17          6          0          0          1          0
-    ##               GSM1281306 GSM1281307
-    ## 0610007P14Rik        105         22
-    ## 0610009B22Rik         15         26
-    ## 0610009L18Rik          0          9
-    ## 0610009O20Rik         24         29
-    ## 0610010F05Rik        108        163
-    ## 0610010K14Rik          0          7
+    ##                    GSM1281300 GSM1281301 GSM1281302 GSM1281303 GSM1281304
+    ## ENSMUSG00000000001       1015        603        561        549        425
+    ## ENSMUSG00000000003          0          0          0          0          0
+    ## ENSMUSG00000000028        109         34          0         14          9
+    ## ENSMUSG00000000031          0         18          0          0          0
+    ## ENSMUSG00000000037          0          0          0          0          0
+    ## ENSMUSG00000000049          0          0          0          0          0
+    ##                    GSM1281305 GSM1281306 GSM1281307
+    ## ENSMUSG00000000001        853        407        479
+    ## ENSMUSG00000000003          0          0          0
+    ## ENSMUSG00000000028        165          0         15
+    ## ENSMUSG00000000031          0          0          0
+    ## ENSMUSG00000000037          0          0          0
+    ## ENSMUSG00000000049          0          0          0
 
 The available gene annotations are also filled in:
 
 ``` r
 head(fData(es))
 ```
 
-    ##                        ENSEMBLID   Gene Symbol
-    ## 0610007P14Rik            missing 0610007P14Rik
-    ## 0610009B22Rik ENSMUSG00000007777 0610009B22Rik
-    ## 0610009L18Rik ENSMUSG00000043644 0610009L18Rik
-    ## 0610009O20Rik            missing 0610009O20Rik
-    ## 0610010F05Rik ENSMUSG00000042208 0610010F05Rik
-    ## 0610010K14Rik ENSMUSG00000020831 0610010K14Rik
+    ##                    Gene symbol          ENSEMBLID
+    ## ENSMUSG00000000001       Gnai3 ENSMUSG00000000001
+    ## ENSMUSG00000000003        Pbsn ENSMUSG00000000003
+    ## ENSMUSG00000000028       Cdc45 ENSMUSG00000000028
+    ## ENSMUSG00000000031         H19 ENSMUSG00000000031
+    ## ENSMUSG00000000037       Scml2 ENSMUSG00000000037
+    ## ENSMUSG00000000049        Apoh ENSMUSG00000000049
diff --git a/README.rmd b/README.rmd
@@ -63,10 +63,10 @@ head(exprs(es))
 
 Function loadCountsFromHSDS returns an ExpressionSet with the expression matrix -- the second exprs(es) contains an expression matrix.
 
-The remote repository URL is '<https://ctlab.itmo.ru/hsds/?domain=/counts>'.
+The remote repository URL is '<https://alserglab.wustl.edu/hsds/?domain=/counts>'.
 
 ```{r}
-url <- 'https://ctlab.itmo.ru/hsds/?domain=/counts'
+url <- 'https://alserglab.wustl.edu/hsds/?domain=/counts'
 es <- loadCountsFromHSDS(es, url)
 head(exprs(es))
 ```

diff --git a/man/getHSDSFileList.Rd b/man/getHSDSFileList.Rd
diff --git a/man/loadCountsFromH5FileHSDS.Rd b/man/loadCountsFromH5FileHSDS.Rd
diff --git a/man/loadCountsFromHSDS.Rd b/man/loadCountsFromHSDS.Rd
diff --git a/tests/testthat/test-loadCountsFromH5file.R b/tests/testthat/test-loadCountsFromH5file.R
@@ -1,7 +1,7 @@
 library(GEOquery)
 
 test_that("loadCountsFromHSDS works correctly", {
-  url <- "https://ctlab.itmo.ru/hsds/?domain=/counts"
+  url <- "https://alserglab.wustl.edu/hsds/?domain=/counts"
   ess <- getGEO("GSE85653", AnnotGPL = TRUE)
   es <- ess[[1]]
   es <- loadCountsFromHSDS(es, url)
@@ -21,7 +21,7 @@ test_that("loadCountsFromHSDS works correctly", {
 
 
 test_that("loadCountsFromHSDS returns the same ExpressionSet, if it contains counts matrix", {
-  url <- "https://ctlab.itmo.ru/hsds/?domain=/counts"
+  url <- "https://alserglab.wustl.edu/hsds/?domain=/counts"
   ess <- getGEO("GSE10010")
   es1 <- ess[[1]]
   es2 <- loadCountsFromHSDS(es1, url)
@@ -31,7 +31,7 @@ test_that("loadCountsFromHSDS returns the same ExpressionSet, if it contains cou
 
 
 test_that("loadCountsFromH5FileHSDS works without metadata params", {
-  url <- "https://ctlab.itmo.ru/hsds/?domain=/counts"
+  url <- "https://alserglab.wustl.edu/hsds/?domain=/counts"
   file <- 'archs4/Arabidopsis_thaliana_count_matrix.h5'
   ess <- getGEO("GSE85653", AnnotGPL = TRUE)
   es <- ess[[1]]

diff --git a/vignettes/phantasusLite-tutorial.Rmd b/vignettes/phantasusLite-tutorial.Rmd
@@ -60,11 +60,11 @@ RNA-seq dataset from GEO do not contain the expression matrix, thus `exprs(es)`
 head(exprs(es))
 ```
 
-However, a number of precomputed gene count tables are available at HSDS server  '<https://ctlab.itmo.ru/hsds/>'. It features HDF5 files with counts
+However, a number of precomputed gene count tables are available at HSDS server  '<https://alserglab.wustl.edu/hsds/>'. It features HDF5 files with counts
 from ARCHS4 and DEE2 projects:
 
 ```{r}
-url <- 'https://ctlab.itmo.ru/hsds/?domain=/counts'
+url <- 'https://alserglab.wustl.edu/hsds/?domain=/counts'
 getHSDSFileList(url)
 ```
 
@@ -92,6 +92,12 @@ The counts are different from the previous values as ARCHS4 counts were used --
 preproc(experimentData(es))$gene_counts_source
 ```
 
+Further, gene symbols are also imported from ARCHS4 database and are available as feature data:
+```{r}
+head(fData(es))
+```
+
+
 # Inferring sample groups
 
 For some of the GEO datasets, such as GSE53053, the sample annotation is not fully available.