feat: precompute user.sigdb connectivity score at pgxcompute

bigomics · Oct 18, 2023 · f30e6eb · f30e6eb
1 parent d4f6474
commit f30e6eb
Show file tree

Hide file tree

Showing 3 changed files with 27 additions and 14 deletions.
diff --git a/NAMESPACE b/NAMESPACE
@@ -115,6 +115,7 @@ export(makeContrastsFromLabelMatrix)
 export(makeDirectContrasts)
 export(makeDirectContrasts000)
 export(makeFullContrasts)
+export(mat2gmt)
 export(matGroupMeans)
 export(merge_sparse_matrix)
 export(mixHivePlot)
@@ -171,7 +172,6 @@ export(pgx.contrastScatter)
 export(pgx.correlateSignatureH5)
 export(pgx.countNormalization)
 export(pgx.createComboDrugAnnot)
-export(pgx.createCreedsSigDB)
 export(pgx.createFromFiles)
 export(pgx.createOmicsGraph)
 export(pgx.createPGX)
@@ -252,10 +252,11 @@ export(pgx.superBatchCorrect)
 export(pgx.survivalVariableImportance)
 export(pgx.testPhenoCorrelation)
 export(pgx.testTCGAsurvival)
-export(pgx.updateInfoPGX)
 export(pgx.variableImportance)
 export(pgx.violinPlot)
 export(pgx.wgcna)
+export(pgxinfo.add)
+export(pgxinfo.delete)
 export(pgxinfo.needUpdate)
 export(pgxinfo.read)
 export(pgxinfo.updateDatasetFolder)

diff --git a/R/compute2-extra.R b/R/compute2-extra.R
@@ -18,7 +18,7 @@
 compute_extra <- function(pgx, extra = c(
                             "meta.go", "infer", "deconv", "drugs", ## "graph",
                             "connectivity", "wordcloud", "wgcna"
-                          ), sigdb = NULL, libx.dir = NULL) {
+                          ), sigdb = NULL, libx.dir = NULL, pgx.dir = NULL) {
   timings <- c()
 
   if (length(extra) == 0) {
@@ -119,25 +119,34 @@ compute_extra <- function(pgx, extra = c(
     message("<<< done!")
   }
 
-  # THIS REQUIRES libx.dir TO BE SET TO FIND sigdb-.h5 FILES
+  # This requires libx.dir to be set (or sigdb passed) to find sigdb-.h5 files
   if ("connectivity" %in% extra) {
     # try to find sigdb in libx dir if not specified
-    if (!is.null(libx.dir) || !is.null(sigdb)) {
+    if (!is.null(libx.dir) || !is.null(pgx.dir) || !is.null(sigdb)) {
       message(">>> Computing connectivity scores...")
       if (is.null(sigdb)) {
-        sigdb <- dir(file.path(libx.dir, "sigdb"), pattern = "h5$", full.names = TRUE)
+        sigdb <- NULL
+        if (!is.null(pgx.dir) ) {
+          ## make sure h5 file is up-to-date
+          pgxinfo.updateDatasetFolder(pgx.dir, force=FALSE, update.sigdb=TRUE)
+          user.sigdb <- file.path(pgx.dir, "datasets-sigdb.h5")
+          sigdb <- c(sigdb, user.sigdb)
+        }
+        if (!is.null(libx.dir) ) {
+          libx.sigdb <- dir(file.path(libx.dir, "sigdb"), pattern = "h5$", full.names = TRUE)
+          sigdb <- c(sigdb, libx.sigdb)
+        }
       }
-
+      
       db <- sigdb[1]
       for (db in sigdb) {
         if (file.exists(db)) {
           message("computing connectivity scores for ", db)
-          ## in memory for many comparisons
-          meta <- pgx.getMetaFoldChangeMatrix(pgx, what = "meta")
           tt <- system.time({
             scores <- pgx.computeConnectivityScores(
-              pgx, db,
-              ntop = 1000,
+              pgx,
+              db,
+              ntop = 200,
               contrasts = NULL,
               remove.le = TRUE
             )

diff --git a/R/pgx-compute.R b/R/pgx-compute.R
@@ -579,7 +579,8 @@ pgx.computePGX <- function(pgx,
                              "meta.go", "infer", "deconv", "drugs",
                              "connectivity", "wordcloud", "wgcna"
                            )[c(1, 2)],
-                           libx.dir = NULL,
+                           pgx.dir = NULL,
+                           libx.dir = NULL,                           
                            progress = NULL) {
   ## ======================================================================
   ## ======================================================================
@@ -643,14 +644,16 @@ pgx.computePGX <- function(pgx,
 
   if (do.cluster) {
     message("[pgx.computePGX] clustering genes...")
-    pgx <- pgx.clusterGenes(pgx, methods = "umap", dims = c(2, 3), level = "geneset") ## gsetX not ready!!
+    ## gsetX was not ready before!!
+    pgx <- pgx.clusterGenes(pgx, methods = "umap", dims = c(2, 3), level = "geneset") 
   }
 
 
   ## ------------------ extra analyses ---------------------
   if (!is.null(progress)) progress$inc(0.3, detail = "extra modules")
   message("[pgx.computePGX] computing extra modules...")
-  pgx <- compute_extra(pgx, extra = extra.methods, libx.dir = libx.dir)
+  pgx <- compute_extra(pgx, extra = extra.methods,
+    pgx.dir = pgx.dir, libx.dir = libx.dir)
 
   message("[pgx.computePGX] done!")