From 7fc9a093c689702f0561fb12027b196693aca2d5 Mon Sep 17 00:00:00 2001
From: Guangchuang Yu <guangchuangyu@gmail.com>
Date: Thu, 15 Aug 2024 17:47:14 +0800
Subject: [PATCH] update mh

---
 mh-mapping.R | 31 ++++++++++++++++++++++++++++++-
 1 file changed, 30 insertions(+), 1 deletion(-)

diff --git a/mh-mapping.R b/mh-mapping.R
index 6acd730..4eb8019 100644
--- a/mh-mapping.R
+++ b/mh-mapping.R
@@ -6,15 +6,44 @@
 #
 
 
+read.sssom <- function(file) {
+    read.delim(file, comment.char = "#")
+}
+
+
 
 dir <- 'mh_mapping_initiative/mappings'
 
 
+hpo2do <- read.sssom(file.path(dir, 'hp_doid_pistoia.sssom.tsv'))
+    
+hpo2do <- hpo2do[, c(1,3)] |>
+    setNames(c("HPO", "DO")) |>
+    unique()
+
+dim(hpo2do)
+head(hpo2do)    
+
+
+hpo2omim <- read.sssom('phenotype.hpoa')[, c("hpo_id", "database_id")] |>
+    setNames(c("hpo_id", "omim_id"))
+do2omim <- read.sssom("OMIMinDO.tsv")[, c("id", "xrefs")] |>
+    setNames(c("do_id", "omim_id"))
+do2omim[,2] <- sprintf("O%s", do2omim[,2])
+head(do2omim)
+
+# too large, the relationship may not true
+hpo2do2 <- merge(hpo2omim, do2omim, by='omim_id')[,c("hpo_id", "do_id")] |> unique()
+
+
+
+
+
 read_mpo2hpo <- function(dir) {
     ff <- list.files(path=dir, pattern="^mp_hp", full.names=TRUE)
 
     res <- lapply(ff, function(f) {
-            x <- read.delim(f, comment.char="#")
+            x <- read.sssom(f)
             x <- x[, c("subject_id", "object_id")]
             x <- x[x[,1] != "sssom:NoTermFound", ]
             x <- x[x[,2] != "sssom:NoTermFound", ]