From 0c429a37eecab4ec6beff207ccfa91b913c564ed Mon Sep 17 00:00:00 2001
From: Christopher Barrington <christopher.barrington@crick.ac.uk>
Date: Tue, 12 Oct 2021 13:23:47 +0100
Subject: [PATCH] added read_msigdb

---
 DESCRIPTION        |  1 +
 NAMESPACE          |  5 +++++
 R/babs.R           | 38 ++++++++++++++++++++++++++++++++++++++
 man/read_msigdb.Rd | 31 +++++++++++++++++++++++++++++++
 4 files changed, 75 insertions(+)
 create mode 100644 man/read_msigdb.Rd

diff --git a/DESCRIPTION b/DESCRIPTION
index f1a9c1b..d25e495 100644
--- a/DESCRIPTION
+++ b/DESCRIPTION
@@ -23,6 +23,7 @@ Imports:
   biomaRt,
   datarepository,
   devtools,
+  fs,
   gtools,
   kableExtra,
   plyr,
diff --git a/NAMESPACE b/NAMESPACE
index e3e8f30..a49cc32 100644
--- a/NAMESPACE
+++ b/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)
diff --git a/R/babs.R b/R/babs.R
index 8bad149..1efb61a 100644
--- a/R/babs.R
+++ b/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(.)])
+}
diff --git a/man/read_msigdb.Rd b/man/read_msigdb.Rd
new file mode 100644
index 0000000..41c4731
--- /dev/null
+++ b/man/read_msigdb.Rd
@@ -0,0 +1,31 @@
+% Generated by roxygen2: do not edit by hand
+% Please edit documentation in R/babs.R
+\name{read_msigdb}
+\alias{read_msigdb}
+\title{Read files from my MSigDB cache}
+\usage{
+read_msigdb(
+  collection = "h.all",
+  pathways = NULL,
+  version = "7.4",
+  identifier = "symbols",
+  dbroot = "/camp/stp/babs/working/barrinc/db/msigdb"
+)
+}
+\arguments{
+\item{collection}{Name of the collection}
+
+\item{pathways}{Character vector of pathways in \code{collection} to keep}
+
+\item{version}{Release version}
+
+\item{identifier}{Identifier type: \code{symbols} or \code{entrez} (\code{entrez} may not exist!!)}
+
+\item{dbroot}{Path to the root of the MSigDB files}
+}
+\value{
+A list of genes in pathways, with pathway name as the key and a character vector of gene identifiers (\code{identifer}).
+}
+\description{
+Reads an MSigDB \code{gmt} file and returns a list of genes in pathways.
+}