added read_msigdb

DESCRIPTION

@@ -23,6 +23,7 @@ Imports:
   biomaRt,
   datarepository,
   devtools,
+  fs,
   gtools,
   kableExtra,
   plyr,

NAMESPACE

@@ -23,6 +23,7 @@ export(open_x11)
 export(preferred_choice)
 export(print_object_size)
 export(read_dotbabs)
+export(read_msigdb)
 export(resize_and_show)
 export(show_newpage_grid)
 export(wideScreen)
@@ -42,10 +43,12 @@ importFrom(devtools,session_info)
 importFrom(dplyr,mutate)
 importFrom(dplyr,n)
 importFrom(dplyr,select)
+importFrom(fs,path)
 importFrom(kableExtra,kable)
 importFrom(magrittr,"%<>%")
 importFrom(magrittr,"%>%")
 importFrom(magrittr,extract)
+importFrom(plyr,dlply)
 importFrom(plyr,llply)
 importFrom(purrr,discard)
 importFrom(purrr,keep)
@@ -62,9 +65,11 @@ importFrom(stringr,str_pad)
 importFrom(stringr,str_remove)
 importFrom(stringr,str_replace_all)
 importFrom(stringr,str_split)
+importFrom(stringr,str_subset)
 importFrom(stringr,str_to_lower)
 importFrom(stringr,str_to_upper)
 importFrom(tibble,deframe)
 importFrom(tibble,is_tibble)
+importFrom(utils,count.fields)
 importFrom(utils,page)
 importFrom(yaml,read_yaml)

R/babs.R

@@ -28,3 +28,41 @@ get_project_root <- function()
     unlist() %>%
     head(n=10) %>%
     str_c(collapse='/')
+
+#' Read files from my MSigDB cache
+#' 
+#' Reads an MSigDB `gmt` file and returns a list of genes in pathways.
+#' 
+#' @param collection Name of the collection
+#' @param pathways Character vector of pathways in `collection` to keep
+#' @param version Release version
+#' @param identifier Identifier type: `symbols` or `entrez` (`entrez` may not exist!!)
+#' @param dbroot Path to the root of the MSigDB files
+#' 
+#' @return
+#' A list of genes in pathways, with pathway name as the key and a character vector of gene identifiers (`identifer`).
+#' 
+#' @importFrom fs path
+#' @importFrom dplyr select
+#' @importFrom plyr dlply
+#' @importFrom purrr when
+#' @importFrom stringr str_c str_subset
+#' @importFrom utils count.fields
+
+#' @export
+#' 
+read_msigdb <- function(collection='h.all', pathways=NULL, version='7.4', identifier='symbols', dbroot='/camp/stp/babs/working/barrinc/db/msigdb') {
+
+  sprintf(fmt='%s/%s.v%s.%s.gmt',dbroot,  collection, version, identifier) %>%
+    when(!file.exists(.)~stop('MSigDB gmt file does not exist!', call.=FALSE),
+         TRUE~.) %>%
+    {list(path=., ncol={count.fields(., sep='\t') %>% max()})} %>%
+    {read.table(file=.$path, sep='\t', fill=TRUE, header=FALSE, col.names=c('pathway', 'url', str_c('id.', 1:.$ncol)))} %>%
+    dlply(~pathway, function(x)
+      select(x, starts_with('id.')) %>%
+        unlist(use.names=FALSE) %>%
+        na.omit() %>%
+        str_subset('^$', negate=TRUE)) %>%
+    when(!is.null(pathways)~.[pathways],
+         TRUE~.[names(.)])
+}