Merge pull request #22 from YuLab-SMU/devel

merge

DESCRIPTION

@@ -1,7 +1,7 @@
 Package: DOSE
 Type: Package
 Title: Disease Ontology Semantic and Enrichment analysis
-Version: 3.27.3
+Version: 3.29.1
 Authors@R: c( person(given = "Guangchuang", family = "Yu",        email = "guangchuangyu@gmail.com",   role = c("aut", "cre")),
               person(given = "Li-Gen",      family = "Wang",      email = "reeganwang020@gmail.com",   role = "ctb"),
               person(given = "Vladislav",   family = "Petyuk",    email = "petyuk@gmail.com",          role = "ctb"),
@@ -29,15 +29,16 @@ Imports:
     reshape2,
     stats,
     utils,
-    MPO.db,
-    HPO.db
+    yulab.utils
 Suggests:
     prettydoc,
     clusterProfiler,
     gson (>= 0.0.5),
     knitr,
     rmarkdown,
     org.Hs.eg.db,
+    HPO.db,
+    MPO.db,
     testthat
 VignetteBuilder: knitr
 ByteCompile: true

Makefile

@@ -1,7 +1,7 @@
 PKGNAME := $(shell sed -n "s/Package: *\([^ ]*\)/\1/p" DESCRIPTION)
 PKGVERS := $(shell sed -n "s/Version: *\([^ ]*\)/\1/p" DESCRIPTION)
 PKGSRC  := $(shell basename `pwd`)
-BIOCVER := RELEASE_3_17
+BIOCVER := RELEASE_3_18
 
 all: rd check clean
 

NAMESPACE

@@ -62,20 +62,8 @@ importFrom(GOSemSim,combineScores)
 importFrom(GOSemSim,load_OrgDb)
 importFrom(GOSemSim,termSim)
 importFrom(HDO.db,HDOANCESTOR)
-importFrom(HDO.db,HDOOFFSPRING)
 importFrom(HDO.db,HDOPARENTS)
 importFrom(HDO.db,HDOTERM)
-importFrom(HPO.db,HPOANCESTOR)
-importFrom(HPO.db,HPOGENE)
-importFrom(HPO.db,HPOOFFSPRING)
-importFrom(HPO.db,HPOPARENTS)
-importFrom(HPO.db,HPOTERM)
-importFrom(MPO.db,MPOANCESTOR)
-importFrom(MPO.db,MPOMGIDO)
-importFrom(MPO.db,MPOMPMGI)
-importFrom(MPO.db,MPOOFFSPRING)
-importFrom(MPO.db,MPOPARENTS)
-importFrom(MPO.db,MPOTERM)
 importFrom(fgsea,fgsea)
 importFrom(ggplot2,aes)
 importFrom(ggplot2,element_blank)
@@ -104,9 +92,12 @@ importFrom(utils,head)
 importFrom(utils,packageDescription)
 importFrom(utils,read.delim)
 importFrom(utils,setTxtProgressBar)
+importFrom(utils,stack)
 importFrom(utils,str)
 importFrom(utils,tail)
 importFrom(utils,txtProgressBar)
+importFrom(yulab.utils,check_pkg)
+importFrom(yulab.utils,get_fun_from_pkg)
 importMethodsFrom(AnnotationDbi,columns)
 importMethodsFrom(AnnotationDbi,exists)
 importMethodsFrom(AnnotationDbi,get)

NEWS.md

@@ -1,3 +1,11 @@
+# DOSE 3.29.1
+
++ mv 'MPO.db' and 'HPO.db' from 'Imports' to 'Suggests' and fixed bugs (2023-11-18, Sat)
+
+# DOSE 3.28.0
+
++ Bioconductor RELEASE_3_18 (2023-10-25, Wed)
+
 # DOSE 3.27.3
 
 + update `TERM2NAME()` to return term if corresponding name not found. (2023-10-09, Mon)

R/enrichDisease.R

@@ -55,3 +55,83 @@ enrichDisease <- function(gene,
     }
     return(res)
 }
+
+##' @importFrom yulab.utils check_pkg
+##' @importFrom yulab.utils get_fun_from_pkg
+##' @importFrom utils stack
+get_ont_info <- function(ontology) {
+    ## selected columns of `genemap`
+    cols <- c(2, 1)
+    if (ontology == "HDO" || ontology == "DO") {
+        check_pkg("HDO.db")
+        # not used in current version
+    } else if (ontology == "HPO") {
+        check_pkg("HPO.db")
+        genemap <- get_fun_from_pkg("HPO.db", "HPOGENE")
+        ancmap <- get_fun_from_pkg("HPO.db", "HPOANCESTOR")
+        termmap <- get_fun_from_pkg("HPO.db", "HPOTERM")
+    } else if (ontology == "MPO") {
+        check_pkg("MPO.db")
+        genemap <- get_fun_from_pkg("MPO.db", "MPOMPMGI")
+        ancmap <- get_fun_from_pkg("MPO.db", "MPOANCESTOR")
+        termmap <- get_fun_from_pkg("MPO.db", "MPOTERM")
+    } else if (ontology == "MDO") {
+        check_pkg("MPO.db")
+        genemap <- get_fun_from_pkg("MPO.db", "MPGMGIDO")
+        cols <- c(1, 2)
+
+        check_pkg("HPO.db")
+        ancmap <- get_fun_from_pkg("HPO.db", "HPOANCESTOR")
+        termmap <- get_fun_from_pkg("HPO.db", "HPOTERM")
+    }
+    # toTable(genemap)[, cols]
+    res <- list(genemap = genemap,
+            cols = cols,
+            ancmap = ancmap,
+            termmap = termmap
+        )
+}
+
+get_dose_data <- function(ontology = "HPO") {
+    if (!exists(".DOSEenv")) .initial()
+    .DOSEEnv <- get(".DOSEEnv", envir = .GlobalEnv)
+    .env <- sprintf(".%s_DOSE_Env", ontology)
+    if (exists(.env, envir=.DOSEEnv)) {
+        res <- get(.env, envir = .DOSEEnv)
+        return(res)
+    }
+
+    assign(.env, new.env(), envir = .DOSEEnv)
+    ret_env <- get(.env, envir = .DOSEEnv)
+
+    ont_info <- get_ont_info(ontology)
+    eg2term <- toTable(ont_info$genemap)[, ont_info$cols]
+
+    TERMS <- names(as.list(ont_info$ancmap))
+    i <- eg2term[,2] %in% TERMS
+    eg2term <- eg2term[i,]
+    # TERM2EG <- split(eg2term[,1], eg2term[,2])
+    EG2TERM <- split(eg2term[,2], eg2term[,1])
+
+    EG2ALLTERM <- lapply(EG2TERM,
+                       function(i) {
+                           ans <- unlist(mget(i, ont_info$ancmap))
+                           ans <- ans[ !is.na(ans) ]
+                           ans <- c(i, ans)
+                           ans <- unique(ans)
+                           return(ans)
+                       })
+
+    EG2ALLTERM.df <- unique(stack(EG2ALLTERM)[, c(2,1)])
+    TERM2ALLEG <- split(EG2ALLTERM.df[,1], EG2ALLTERM.df[,2])
+
+    PATH2NAME.df <- unique(toTable(ont_info$termmap))
+    PATH2NAME <- setNames(PATH2NAME.df[,2], PATH2NAME.df[,1])        
+
+    assign("EXTID2PATHID", EG2ALLTERM, envir = ret_env)
+    assign("PATHID2EXTID", TERM2ALLEG, envir = ret_env)
+    assign("PATHID2NAME", PATH2NAME, envir = ret_env)
+
+    return(ret_env)    
+}
+

R/enrichHPO.r

@@ -34,56 +34,8 @@ enrichHPO <- function(gene,
 
 
 #' Get HPO data
-#' @importFrom HPO.db HPOANCESTOR
-#' @importFrom HPO.db HPOPARENTS
-#' @importFrom HPO.db HPOTERM
-#' @importFrom HPO.db HPOGENE
 #' @noRd 
 get_HPO_data <- function() {
-    if (!exists(".DOSEenv")) .initial()
-    .DOSEEnv <- get(".DOSEEnv", envir = .GlobalEnv)
-    if (!exists(".HPO_DOSE_Env", envir=.DOSEEnv)) {
-        assign(".HPO_DOSE_Env", new.env(), envir = .DOSEEnv)
-        .HPO_DOSE_Env <- get(".HPO_DOSE_Env", envir = .DOSEEnv)
-        ## get EXTID2PATHID gene_doid
-        # (1) get eg.do
-        # colnames(anno_old) <- c("eg", "doid")
-        eg.do <- toTable(HPOGENE)[, c(2,1)]
-        colnames(eg.do) <- c("eg", "doid")
-        # (2) DOSE:::rebuildAnnoData.internal(eg.do)
-        DO2EG <- with(eg.do, split(as.character(eg), as.character(doid)))
-        DOTERMs <- names(as.list(HPOANCESTOR))
-        idx <- names(DO2EG) %in% DOTERMs
-        DO2EG <- DO2EG[idx]
-        DO2EG <- lapply(DO2EG, function(i) unique(i))
-        EG2DO <- with(eg.do, split(as.character(doid), as.character(eg)))
-        EG2DO <- lapply(EG2DO, function(i) unique(i[ i %in% DOTERMs ]))
-        i <- unlist(lapply(EG2DO, function(i) length(i) != 0))
-        EG2DO <- EG2DO[i]        
-        EG2ALLDO <- lapply(EG2DO,
-                           function(i) {
-                               ans <- unlist(mget(i, HPOANCESTOR))
-                               ans <- ans[ !is.na(ans) ]
-                               ans <- c(i, ans)
-                               ans <- unique(ans)
-                               return(ans)
-                           })      
-        len <- lapply(EG2ALLDO,length)
-        EG2ALLDO.df <- data.frame(EG=rep(names(EG2ALLDO), times=len),
-                                  DO=unlist(EG2ALLDO))
-        DO <- NULL ## satisfy code tools
-        DO2ALLEG <- with(EG2ALLDO.df, split(as.character(EG), as.character(DO)))
-        DO2ALLEG <- lapply(DO2ALLEG, unique)
-        PATH2NAME.df <- toTable(HPOTERM)
-        colnames(PATH2NAME.df) <- c("doid", "term")
-        PATH2NAME.df <- unique(PATH2NAME.df)
-        PATH2NAME <- PATH2NAME.df[,2]
-        names(PATH2NAME) <- PATH2NAME.df[,1]        
-    }
-    assign("EXTID2PATHID", EG2ALLDO, envir = .HPO_DOSE_Env)
-    assign("PATHID2EXTID", DO2ALLEG, envir = .HPO_DOSE_Env)
-    assign("PATHID2NAME", PATH2NAME, envir = .HPO_DOSE_Env)
-
-    get(".HPO_DOSE_Env", envir = .DOSEEnv)
+    get_dose_data("HPO")
 }
 

R/enrichMPO.r

@@ -35,96 +35,9 @@ enrichMPO <- function(gene,
 
 #' Get MPO data
 #' @param ont one of "DO" and "MPO".
-#' @importFrom MPO.db MPOMGIDO
-#' @importFrom MPO.db MPOANCESTOR
-#' @importFrom MPO.db MPOPARENTS
-#' @importFrom MPO.db MPOMPMGI
-#' @importFrom MPO.db MPOTERM
 #' @noRd 
 get_MPO_data <- function(ont = "DO") {
-    if (!exists(".DOSEenv")) .initial()
-    .DOSEEnv <- get(".DOSEEnv", envir = .GlobalEnv)
-    if (!exists(".MPO_DOSE_Env", envir=.DOSEEnv)) {
-        assign(".MPO_DOSE_Env", new.env(), envir = .DOSEEnv)
-        .MPO_DOSE_Env <- get(".MPO_DOSE_Env", envir = .DOSEEnv)
-        if (ont == "DO") {
-            ## get EXTID2PATHID gene_doid
-            # (1) get eg.do
-            # colnames(anno_old) <- c("eg", "doid")
-            eg.do <- toTable(MPOMGIDO)
-            colnames(eg.do) <- c("eg", "doid")
-            # (2) DOSE:::rebuildAnnoData.internal(eg.do)
-            DO2EG <- with(eg.do, split(as.character(eg), as.character(doid)))
-            DOTERMs <- names(as.list(HDOANCESTOR))
-            idx <- names(DO2EG) %in% DOTERMs
-            DO2EG <- DO2EG[idx]
-            DO2EG <- lapply(DO2EG, function(i) unique(i))
-            EG2DO <- with(eg.do, split(as.character(doid), as.character(eg)))
-            EG2DO <- lapply(EG2DO, function(i) unique(i[ i %in% DOTERMs ]))
-            i <- unlist(lapply(EG2DO, function(i) length(i) != 0))
-            EG2DO <- EG2DO[i]        
-            EG2ALLDO <- lapply(EG2DO,
-                               function(i) {
-                                   ans <- unlist(mget(i, HDOANCESTOR))
-                                   ans <- ans[ !is.na(ans) ]
-                                   ans <- c(i, ans)
-                                   ans <- unique(ans)
-                                   return(ans)
-                               })      
-            len <- lapply(EG2ALLDO,length)
-            EG2ALLDO.df <- data.frame(EG=rep(names(EG2ALLDO), times=len),
-                                      DO=unlist(EG2ALLDO))
-            DO <- NULL ## satisfy code tools
-            DO2ALLEG <- with(EG2ALLDO.df, split(as.character(EG), as.character(DO)))
-            DO2ALLEG <- lapply(DO2ALLEG, unique)
-
-            PATH2NAME.df <- toTable(HDOTERM)
-            # PATH2NAME.df <- PATH2NAME.df[, c("do_id", "Term")]
-            PATH2NAME.df <- PATH2NAME.df[, c("doid", "term")]
-            PATH2NAME.df <- unique(PATH2NAME.df)
-            PATH2NAME <- PATH2NAME.df[,2]
-            names(PATH2NAME) <- PATH2NAME.df[,1]
-
-        } else {
-            ## get EXTID2PATHID gene_doid
-            # (1) get eg.do
-            # colnames(anno_old) <- c("eg", "doid")
-            eg.do <- toTable(MPOMPMGI)[, c(2,1)]
-            colnames(eg.do) <- c("eg", "doid")
-            # (2) DOSE:::rebuildAnnoData.internal(eg.do)
-            DO2EG <- with(eg.do, split(as.character(eg), as.character(doid)))
-            DOTERMs <- names(as.list(MPOANCESTOR))
-            idx <- names(DO2EG) %in% DOTERMs
-            DO2EG <- DO2EG[idx]
-            DO2EG <- lapply(DO2EG, function(i) unique(i))
-            EG2DO <- with(eg.do, split(as.character(doid), as.character(eg)))
-            EG2DO <- lapply(EG2DO, function(i) unique(i[ i %in% DOTERMs ]))
-            i <- unlist(lapply(EG2DO, function(i) length(i) != 0))
-            EG2DO <- EG2DO[i]        
-            EG2ALLDO <- lapply(EG2DO,
-                               function(i) {
-                                   ans <- unlist(mget(i, MPOANCESTOR))
-                                   ans <- ans[ !is.na(ans) ]
-                                   ans <- c(i, ans)
-                                   ans <- unique(ans)
-                                   return(ans)
-                               })      
-            len <- lapply(EG2ALLDO,length)
-            EG2ALLDO.df <- data.frame(EG=rep(names(EG2ALLDO), times=len),
-                                      DO=unlist(EG2ALLDO))
-            DO <- NULL ## satisfy code tools
-            DO2ALLEG <- with(EG2ALLDO.df, split(as.character(EG), as.character(DO)))
-            DO2ALLEG <- lapply(DO2ALLEG, unique)
-            PATH2NAME.df <- toTable(MPOTERM)
-            colnames(PATH2NAME.df) <- c("doid", "term")
-            PATH2NAME.df <- unique(PATH2NAME.df)
-            PATH2NAME <- PATH2NAME.df[,2]
-            names(PATH2NAME) <- PATH2NAME.df[,1]        
-        }
-        assign("EXTID2PATHID", EG2ALLDO, envir = .MPO_DOSE_Env)
-        assign("PATHID2EXTID", DO2ALLEG, envir = .MPO_DOSE_Env)
-        assign("PATHID2NAME", PATH2NAME, envir = .MPO_DOSE_Env)
-    }
-    get(".MPO_DOSE_Env", envir = .DOSEEnv)
+    if (ont == "DO") ont = "MDO"
+    get_dose_data(ont)
 }