fix: make DECIPHER HI import work again (#90)

varfish-org · Mar 6, 2024 · 506c636 · 506c636
1 parent c3fdc5a
commit 506c636
Show file tree

Hide file tree

Showing 5 changed files with 6 additions and 5 deletions.
diff --git a/Snakefile b/Snakefile
@@ -90,7 +90,7 @@ rule all:
         f"work/download/genes/ctd/{DV.today}/CTD_diseases.tsv.gz",
         f"work/download/do/{DV.today}/omim-unmapped.csv",
         f"work/genes/dbnsfp/{DV.dbnsfp}/genes.tsv.gz",
-        "work/genes/decipher/v3/decipher_hi_prediction.tsv.gz",
+        "work/genes/decipher/v3/decipher_hi_prediction.tsv",
         f"work/genes/ensembl/{DV.ensembl}/ensembl_xlink.tsv",
         f"work/genes/enst_ensg/grch37/{DV.ensembl_37}/enst_ensg.tsv",
         f"work/genes/entrez/{DV.today}/gene_info.jsonl",

diff --git a/rules/output/annonars/genes.smk b/rules/output/annonars/genes.smk
@@ -18,7 +18,7 @@ rule output_annonars_genes:  # -- build annonars genes RocksDB file
         shet="work/genes/shet/2019/shet_weghorn_2019.tsv",
         gtex="work/genes/annonars/gtex_v8/genes_tpm.jsonl.gz",
         domino="work/genes/domino/20190219/domino.tsv",
-        decipher_hi="work/genes/decipher/v3/decipher_hi_prediction.tsv.gz",
+        decipher_hi="work/genes/decipher/v3/decipher_hi_prediction.tsv",
     output:
         rocksdb_identity=(
             "output/full/annonars/genes-{v_acmg_sf}+{v_gnomad_constraints}+{v_dbnsfp}+{v_hpo}+{date}+{v_annonars}/"

diff --git a/rules/work/genes/decipher.smk b/rules/work/genes/decipher.smk
@@ -20,8 +20,8 @@ rule genes_decipher_hi_convert:  # -- convert DECIPHER HI predictions to TSV
         hgnc=f"output/full/mehari/genes-xlink-{DV.today}/genes-xlink.tsv",
         bed="work/download/genes/decipher/v3/HI_Predictions_Version3.bed.gz",
     output:
-        tsv="work/genes/decipher/v3/decipher_hi_prediction.tsv.gz",
-        tsv_md5="work/genes/decipher/v3/decipher_hi_prediction.tsv.gz.md5",
+        tsv="work/genes/decipher/v3/decipher_hi_prediction.tsv",
+        tsv_md5="work/genes/decipher/v3/decipher_hi_prediction.tsv.md5",
     shell:
         r"""
         set -x

diff --git a/scripts/genes-integrate-diseases.py b/scripts/genes-integrate-diseases.py
@@ -1210,6 +1210,7 @@ def handle_panelapp(self):  # noqa: C901
                     omim_ids: List[LabeledDisorder] = []
                     orpha_ids: List[LabeledDisorder] = []
                     orpha_disorder: Optional[OrphaDisorder] = None
+                    omim_id = ""  # in case it's unset
                     if m_mondo:
                         mondo_id = f"MONDO:{m_mondo.group(1)}"
                         omim_ids = list(self.mondo_to_omim.get(mondo_id, []))

diff --git a/varfish_db_downloader/versions.py b/varfish_db_downloader/versions.py
@@ -128,7 +128,7 @@ class DataVersions:
     clingen_variant=TODAY,
     ensembl_37="87",
     ensembl_38="109",
-    ensembl="110",
+    ensembl="111",
     today=TODAY,
     dbnsfp="4.5",
     dbscsnv="1.1",