From aacb00aee0b1e5982cf9d3195471992f470e6eb7 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Irene=20L=C3=B3pez?= <irene.lopezs@protonmail.com>
Date: Thu, 19 Oct 2023 11:46:27 +0100
Subject: [PATCH 1/3] fix(gene_index): typo in gene_index output dataset

---
 config/datasets/gcp.yaml | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/config/datasets/gcp.yaml b/config/datasets/gcp.yaml
index 9b7af54a0..578e76b14 100644
--- a/config/datasets/gcp.yaml
+++ b/config/datasets/gcp.yaml
@@ -27,7 +27,7 @@ ld_index_raw_template: gs://gcp-public-data--gnomad/release/2.1.1/ld/gnomad.geno
 ld_matrix_template: gs://gcp-public-data--gnomad/release/2.1.1/ld/gnomad.genomes.r2.1.1.{POP}.common.adj.ld.bm
 
 # Output datasets
-gene_index: ${datasets.outputs}/gene_index/gene_index
+gene_index: ${datasets.outputs}/gene_index
 variant_annotation: ${datasets.outputs}/variant_annotation
 variant_index: ${datasets.outputs}/variant_index
 study_locus: ${datasets.outputs}/study_locus

From 68a280b484d465601238ab5dc4381e80e990b4c8 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Irene=20L=C3=B3pez?= <irene.lopezs@protonmail.com>
Date: Thu, 19 Oct 2023 12:36:38 +0100
Subject: [PATCH 2/3] fix(gene_index): add `approvedName` `approvedSymbol`
 `biotype` to `as_gene_index`

---
 src/otg/datasource/open_targets/target.py | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/src/otg/datasource/open_targets/target.py b/src/otg/datasource/open_targets/target.py
index 3d334d34d..64172a99c 100644
--- a/src/otg/datasource/open_targets/target.py
+++ b/src/otg/datasource/open_targets/target.py
@@ -49,6 +49,9 @@ def as_gene_index(cls: type[GeneIndex], target_index: DataFrame) -> GeneIndex:
         return GeneIndex(
             _df=target_index.select(
                 f.coalesce(f.col("id"), f.lit("unknown")).alias("geneId"),
+                "approvedSymbol",
+                "approvedName",
+                "biotype",
                 f.coalesce(f.col("genomicLocation.chromosome"), f.lit("unknown")).alias(
                     "chromosome"
                 ),

From ec237667faf6d307b8d95693ff85f5c3ffdd999e Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Irene=20L=C3=B3pez?= <irene.lopezs@protonmail.com>
Date: Thu, 19 Oct 2023 12:40:47 +0100
Subject: [PATCH 3/3] fix(gene_index): add `obsoleteSymbols` and simplify
 schema

---
 src/otg/assets/schemas/gene_index.json    | 28 +++++------------------
 src/otg/dataset/gene_index.py             |  2 +-
 src/otg/datasource/open_targets/target.py |  1 +
 3 files changed, 8 insertions(+), 23 deletions(-)

diff --git a/src/otg/assets/schemas/gene_index.json b/src/otg/assets/schemas/gene_index.json
index cd901a995..c8139c20d 100644
--- a/src/otg/assets/schemas/gene_index.json
+++ b/src/otg/assets/schemas/gene_index.json
@@ -32,30 +32,14 @@
       "metadata": {}
     },
     {
+      "metadata": {},
       "name": "obsoleteSymbols",
-      "type": {
-        "type": "array",
-        "elementType": {
-          "type": "struct",
-          "fields": [
-            {
-              "name": "label",
-              "type": "string",
-              "nullable": true,
-              "metadata": {}
-            },
-            {
-              "name": "source",
-              "type": "string",
-              "nullable": true,
-              "metadata": {}
-            }
-          ]
-        },
-        "containsNull": true
-      },
       "nullable": true,
-      "metadata": {}
+      "type": {
+        "containsNull": true,
+        "elementType": "string",
+        "type": "array"
+      }
     },
     {
       "name": "tss",
diff --git a/src/otg/dataset/gene_index.py b/src/otg/dataset/gene_index.py
index 3b67560da..979ef959e 100644
--- a/src/otg/dataset/gene_index.py
+++ b/src/otg/dataset/gene_index.py
@@ -64,7 +64,7 @@ def symbols_lut(self: GeneIndex) -> DataFrame:
         """
         return self.df.select(
             f.explode(
-                f.array_union(f.array("approvedSymbol"), f.col("obsoleteSymbols.label"))
+                f.array_union(f.array("approvedSymbol"), f.col("obsoleteSymbols"))
             ).alias("geneSymbol"),
             "*",
         )
diff --git a/src/otg/datasource/open_targets/target.py b/src/otg/datasource/open_targets/target.py
index 64172a99c..78096c6e3 100644
--- a/src/otg/datasource/open_targets/target.py
+++ b/src/otg/datasource/open_targets/target.py
@@ -52,6 +52,7 @@ def as_gene_index(cls: type[GeneIndex], target_index: DataFrame) -> GeneIndex:
                 "approvedSymbol",
                 "approvedName",
                 "biotype",
+                f.col("obsoleteSymbols.label").alias("obsoleteSymbols"),
                 f.coalesce(f.col("genomicLocation.chromosome"), f.lit("unknown")).alias(
                     "chromosome"
                 ),