wip

varfish-org · Jan 19, 2024 · 3a33bb8 · 3a33bb8
1 parent ce44ee4
commit 3a33bb8
Show file tree

Hide file tree

Showing 7 changed files with 38 additions and 6 deletions.
diff --git a/rules/output/annonars/genes.smk b/rules/output/annonars/genes.smk
@@ -36,6 +36,11 @@ rule output_annonars_genes:  # -- build annonars genes RocksDB file
         v_annonars=RE_VERSION,
     shell:
         r"""
+        if [[ "$(date +%Y%m%d)" != "{wildcards.date}" ]] && [[ "{FORCE_TODAY}" != "True" ]]; then
+            >&2 echo "{wildcards.date} is not today"
+            exit 1
+        fi
+
         annonars gene import \
             --path-out-rocksdb $(dirname {output.rocksdb_identity}) \
             --path-in-acmg {input.acmg_sf} \

diff --git a/rules/output/worker/hgnc.smk b/rules/output/worker/hgnc.smk
@@ -9,6 +9,11 @@ rule output_hgnc_xlink_binary:
         spec_yaml=f"output/full/worker/genes-xlink-{{date}}+{PV.worker}/genes-xlink.spec.yaml",
     shell:
         r"""
+        if [[ "$(date +%Y%m%d)" != "{wildcards.date}" ]] && [[ "{FORCE_TODAY}" != "True" ]]; then
+            >&2 echo "{wildcards.date} is not today"
+            exit 1
+        fi
+
         varfish-server-worker db to-bin \
             --input-type xlink \
             --path-input {input.tsv} \

diff --git a/rules/work/genes/conditions.smk b/rules/work/genes/conditions.smk
@@ -18,6 +18,11 @@ rule work_conditions_integrate:  # integrate conditions file
         jsonl="work/genes/conditions/{v_hpo}+{date}/conditions.jsonl",
     shell:
         r"""
+        if [[ "$(date +%Y%m%d)" != "{wildcards.date}" ]] && [[ "{FORCE_TODAY}" != "True" ]]; then
+            >&2 echo "{wildcards.date} is not today"
+            exit 1
+        fi
+
         MIM2GENE_MEDGEN_PATH="{input.mim2gene_medgen}" \
         GENES_XLINK_PATH="{input.genes_xlink}" \
         HPOA_PATH="{input.hpoa}" \

diff --git a/rules/work/genes/ctd.smk b/rules/work/genes/ctd.smk
@@ -6,5 +6,10 @@ rule genes_ctd_download:  # -- download CTD
         tsv="work/download/genes/ctd/{date}/CTD_diseases.tsv.gz",
     shell:
         """
+        if [[ "$(date +%Y%m%d)" != "{wildcards.date}" ]] && [[ "{FORCE_TODAY}" != "True" ]]; then
+            >&2 echo "{wildcards.date} is not today"
+            exit 1
+        fi
+
         wget -O {output.tsv} https://ctdbase.org/reports/CTD_diseases.tsv.gz
         """
diff --git a/rules/work/genes/do.smk b/rules/work/genes/do.smk
@@ -8,6 +8,11 @@ rule work_disease_ontology_download:  # -- download DO files
         omim_import="work/download/do/{date}/omim_import.obo",
     shell:
         r"""
+        if [[ "$(date +%Y%m%d)" != "{wildcards.date}" ]] && [[ "{FORCE_TODAY}" != "True" ]]; then
+            >&2 echo "{wildcards.date} is not today"
+            exit 1
+        fi
+
         wget -O {output.unmapped_csv} \
             https://github.com/DiseaseOntology/HumanDiseaseOntology/raw/main/src/deprecated/reports/omim-unmapped.csv
         wget -O {output.omimindo_tsv} \

diff --git a/rules/work/genes/omim.smk b/rules/work/genes/omim.smk
@@ -12,8 +12,13 @@ rule genes_omim:  # -- prepare HGNC to OMIM disease mapping
         """
         set -x
 
+        if [[ "$(date +%Y%m%d)" != "{wildcards.date}" ]] && [[ "{FORCE_TODAY}" != "True" ]]; then
+            >&2 echo "{wildcards.date} is not today"
+            exit 1
+        fi
+
         export TMPDIR=$(mktemp -d)
-        # trap "rm -rf $TMPDIR" ERR EXIT
+        trap "rm -rf $TMPDIR" ERR EXIT
 
         head -n 1 {input.mim2gene} | sed -e 's/ /_/g' -e 's/#//g' \
         > $TMPDIR/mim2gene.tsv

diff --git a/rules/work/genes/orphadata.smk b/rules/work/genes/orphadata.smk
@@ -3,13 +3,15 @@
 
 rule genes_orphadata_download:  # -- download orphadatas
     output:
-        jsonl="work/genes/orphadata/{version}/orphadata.jsonl",
-        jsonl_md5="work/genes/orphadata/{version}/orphadata.jsonl.md5",
-        tsv="work/genes/orphadata/{version}/orpha_diseases.tsv",  # xxx
+        jsonl="work/genes/orphadata/{date}/orphadata.jsonl",
+        jsonl_md5="work/genes/orphadata/{date}/orphadata.jsonl.md5",
+        tsv="work/genes/orphadata/{date}/orpha_diseases.tsv",  # xxx
     shell:
         """
-        export TMPDIR=$(mktemp -d)
-        trap "rm -rf $TMPDIR" ERR EXIT
+        if [[ "$(date +%Y%m%d)" != "{wildcards.date}" ]] && [[ "{FORCE_TODAY}" != "True" ]]; then
+            >&2 echo "{wildcards.date} is not today"
+            exit 1
+        fi
 
         python ./scripts/genes-orpha-diseases.py \
         > {output.jsonl}