From 0055a318d4069be8814493392d1bebe6f7fbb57e Mon Sep 17 00:00:00 2001
From: Manuel Holtgrewe <manuel.holtgrewe@bih-charite.de>
Date: Tue, 21 Nov 2023 18:03:59 +0100
Subject: [PATCH] feat: build annonars regions

---
 Snakefile                               |  4 ++
 rules/output/annonars/regions.smk       | 54 +++++++++++++++++++++++++
 rules/output/annonars/regions.spec.yaml | 14 +++++++
 3 files changed, 72 insertions(+)
 create mode 100644 rules/output/annonars/regions.smk
 create mode 100644 rules/output/annonars/regions.spec.yaml

diff --git a/Snakefile b/Snakefile
index c0f1dea..c1ea664 100644
--- a/Snakefile
+++ b/Snakefile
@@ -165,8 +165,11 @@ rule all:
         f"output/full/annonars/gnomad-sv-exomes-grch38-{DV.gnomad_cnv4}+{PV.annonars}/rocksdb/IDENTITY",
         f"output/full/annonars/gnomad-sv-genomes-grch37-{DV.gnomad_sv}+{PV.annonars}/rocksdb/IDENTITY",
         f"output/full/annonars/gnomad-sv-genomes-grch38-{DV.gnomad_sv4}+{PV.annonars}/rocksdb/IDENTITY",
+        # ----- sequence annotation
         f"output/full/annonars/functional-grch37-{DV.refseq_fe_37}+{PV.annonars}/rocksdb/IDENTITY",
         f"output/full/annonars/functional-grch38-{DV.refseq_fe_38}+{PV.annonars}/rocksdb/IDENTITY",
+        f"output/full/annonars/regions-grch37-{DV.today}+{PV.annonars}/rocksdb/IDENTITY",
+        f"output/full/annonars/regions-grch38-{DV.today}+{PV.annonars}/rocksdb/IDENTITY",
         # ----- conservation
         f"output/full/annonars/cons-grch37-{DV.ucsc_cons_37}+{PV.annonars}/rocksdb/IDENTITY",
         f"output/full/annonars/cons-grch38-{DV.ucsc_cons_38}+{PV.annonars}/rocksdb/IDENTITY",
@@ -389,6 +392,7 @@ include: "rules/output/annonars/gnomad_sv.smk"
 include: "rules/output/annonars/helix.smk"
 include: "rules/output/annonars/genes.smk"
 include: "rules/output/annonars/functional.smk"
+include: "rules/output/annonars/regions.smk"
 # ---- worker
 include: "rules/output/worker/patho_mms.smk"
 include: "rules/output/worker/clinvar.smk"
diff --git a/rules/output/annonars/regions.smk b/rules/output/annonars/regions.smk
new file mode 100644
index 0000000..a93ad97
--- /dev/null
+++ b/rules/output/annonars/regions.smk
@@ -0,0 +1,54 @@
+## Rules to create build annonars regions annotation database..
+
+
+rule work_annonars_regions_download:  # -- download clingen regions
+    output:
+        "work/download/clingen/{genome_release}/{today}/ClinGen_region_curation_list_{genome_release}.tsv",
+    shell:
+        r"""
+        if [[ "{wildcards.genome_release}" == "grch38" ]]; then
+            GENOME=GRCh37
+        else
+            GENOME=GRCh38
+        fi
+
+        wget -O {output} \
+            ftp://ftp.clinicalgenome.org/ClinGen_region_curation_list_$GENOME.tsv
+        """
+
+rule output_annonars_regions:  # -- build annonars regions RocksDB file
+    input:
+        "work/download/clingen/{genome_release}/{date}/ClinGen_region_curation_list_{genome_release}.tsv",
+    output:
+        rocksdb_identity=(
+            "output/full/annonars/regions-{genome_release}-{date}+{v_annonars}/"
+            "rocksdb/IDENTITY"
+        ),
+        spec_yaml=(
+            "output/full/annonars/regions-{genome_release}-{date}+{v_annonars}/spec.yaml"
+        ),
+    wildcard_constraints:
+        v_refseq=RE_VERSION,
+        v_annonars=RE_VERSION,
+    shell:
+        r"""
+        if [[ "$(date +%Y%m%d)" != "{wildcards.date}" ]] && [[ "{FORCE_TODAY}" != "True" ]]; then
+            >&2 echo "{wildcards.date} is not today"
+            exit 1
+        fi
+
+        annonars regions import -vvv \
+            --genome-release {wildcards.genome_release} \
+            --path-in-clingen {input} \
+            --path-out-rocksdb $(dirname {output.rocksdb_identity})
+
+        varfish-db-downloader tpl \
+            --template rules/output/annonars/regions.spec.yaml \
+            --value today={TODAY} \
+            \
+            --value version={wildcards.date}+{wildcards.v_annonars} \
+            \
+            --value v_annonars={wildcards.v_annonars} \
+            --value v_downloader={PV.downloader} \
+        > {output.spec_yaml}
+        """
diff --git a/rules/output/annonars/regions.spec.yaml b/rules/output/annonars/regions.spec.yaml
new file mode 100644
index 0000000..88e5206
--- /dev/null
+++ b/rules/output/annonars/regions.spec.yaml
@@ -0,0 +1,14 @@
+dc.identifier: annonars/regions:{{ version }}-{{ genome_release }}
+dc.title: annonars regions annotation RocksDB
+dc.creator: VarFish Developer Teams
+dc.format: application/x-rocksdb
+dc.date: {{ today }}
+x-version: {{ version }}
+x-genome-release: {{ genome_release }}
+dc.description: |
+  RocksDB with region annotation.
+dc.source:
+  - https://search.clinicalgenome.org/kb/gene-dosage
+x-created-from:
+  - name: ClinGen Region Dosage Pathogenicity
+    version: {{ today }}