feat: adjusting output paths (#42)

holtgrewe · web-flow · commit 4b01fe9340bf · 2023-06-14T10:53:12.000Z
diff --git a/Snakefile b/Snakefile
@@ -115,36 +115,27 @@ rule all:
         # ---- frequencies (via annonars)
         f"output/mehari/freqs-grch37-{DV.gnomad_v2}+{DV.gnomad_v2}+{DV.gnomad_mtdna}+{DV.helixmtdb}+{PV.annonars}/rocksdb/IDENTITY",
         f"output/mehari/freqs-grch38-{DV.gnomad_v3}+{DV.gnomad_v2}+{DV.gnomad_mtdna}+{DV.helixmtdb}+{PV.annonars}/rocksdb/IDENTITY",
-        # -- varfish-server-worker data
-        # ---- CADD
-        f"output/worker/annos/seqvars/cadd-grch37-{DV.cadd}+{PV.annonars}/rocksdb/IDENTITY",
-        f"output/worker/annos/seqvars/cadd-grch38-{DV.cadd}+{PV.annonars}/rocksdb/IDENTITY",
-        # ---- dbSNP
-        f"output/worker/annos/seqvars/dbsnp-grch37-{DV.dbsnp}+{PV.annonars}/rocksdb/IDENTITY",
-        f"output/worker/annos/seqvars/dbsnp-grch38-{DV.dbsnp}+{PV.annonars}/rocksdb/IDENTITY",
-        # ---- dbNSFP
-        f"output/worker/annos/seqvars/dbnsfp-grch37-{DV.dbnsfp}a+{PV.annonars}/rocksdb/IDENTITY",
-        f"output/worker/annos/seqvars/dbnsfp-grch38-{DV.dbnsfp}a+{PV.annonars}/rocksdb/IDENTITY",
-        f"output/worker/annos/seqvars/dbnsfp-grch37-{DV.dbnsfp}c+{PV.annonars}/rocksdb/IDENTITY",
-        f"output/worker/annos/seqvars/dbnsfp-grch38-{DV.dbnsfp}c+{PV.annonars}/rocksdb/IDENTITY",
-        # ---- dbscSNV
-        f"output/worker/annos/seqvars/dbscsnv-grch37-{DV.dbscsnv}+{PV.annonars}/rocksdb/IDENTITY",
-        f"output/worker/annos/seqvars/dbscsnv-grch38-{DV.dbscsnv}+{PV.annonars}/rocksdb/IDENTITY",
-        # ---- gnomAD mtDNA
-        f"output/worker/annos/seqvars/gnomad-mtdna-grch37-{DV.gnomad_mtdna}+{PV.annonars}/rocksdb/IDENTITY",
-        f"output/worker/annos/seqvars/gnomad-mtdna-grch38-{DV.gnomad_mtdna}+{PV.annonars}/rocksdb/IDENTITY",
-        # ---- gnomAD exomes
-        f"output/worker/annos/seqvars/gnomad-exomes-grch37-{DV.gnomad_v2}+{PV.annonars}/rocksdb/IDENTITY",
-        f"output/worker/annos/seqvars/gnomad-exomes-grch38-{DV.gnomad_v2}+{PV.annonars}/rocksdb/IDENTITY",
-        # ---- gnomAD genomes
-        f"output/worker/annos/seqvars/gnomad-genomes-grch37-{DV.gnomad_v2}+{PV.annonars}/rocksdb/IDENTITY",
-        f"output/worker/annos/seqvars/gnomad-genomes-grch38-{DV.gnomad_v3}+{PV.annonars}/rocksdb/IDENTITY",
-        # ---- HelixMtDb
-        f"output/worker/annos/seqvars/helixmtdb-grch37-{DV.helixmtdb}+{PV.annonars}/rocksdb/IDENTITY",
-        f"output/worker/annos/seqvars/helixmtdb-grch38-{DV.helixmtdb}+{PV.annonars}/rocksdb/IDENTITY",
-        # ---- UCSC conservation
-        f"output/worker/annos/seqvars/cons-grch37-{DV.ucsc_cons_37}+{PV.annonars}/rocksdb/IDENTITY",
-        f"output/worker/annos/seqvars/cons-grch38-{DV.ucsc_cons_38}+{PV.annonars}/rocksdb/IDENTITY",
+        # ---- annonars data
+        f"output/annonars/cadd-grch37-{DV.cadd}+{PV.annonars}/rocksdb/IDENTITY",
+        f"output/annonars/cadd-grch38-{DV.cadd}+{PV.annonars}/rocksdb/IDENTITY",
+        f"output/annonars/dbsnp-grch37-{DV.dbsnp}+{PV.annonars}/rocksdb/IDENTITY",
+        f"output/annonars/dbsnp-grch38-{DV.dbsnp}+{PV.annonars}/rocksdb/IDENTITY",
+        f"output/annonars/dbnsfp-grch37-{DV.dbnsfp}a+{PV.annonars}/rocksdb/IDENTITY",
+        f"output/annonars/dbnsfp-grch38-{DV.dbnsfp}a+{PV.annonars}/rocksdb/IDENTITY",
+        f"output/annonars/dbnsfp-grch37-{DV.dbnsfp}c+{PV.annonars}/rocksdb/IDENTITY",
+        f"output/annonars/dbnsfp-grch38-{DV.dbnsfp}c+{PV.annonars}/rocksdb/IDENTITY",
+        f"output/annonars/dbscsnv-grch37-{DV.dbscsnv}+{PV.annonars}/rocksdb/IDENTITY",
+        f"output/annonars/dbscsnv-grch38-{DV.dbscsnv}+{PV.annonars}/rocksdb/IDENTITY",
+        f"output/annonars/gnomad-mtdna-grch37-{DV.gnomad_mtdna}+{PV.annonars}/rocksdb/IDENTITY",
+        f"output/annonars/gnomad-mtdna-grch38-{DV.gnomad_mtdna}+{PV.annonars}/rocksdb/IDENTITY",
+        f"output/annonars/gnomad-exomes-grch37-{DV.gnomad_v2}+{PV.annonars}/rocksdb/IDENTITY",
+        f"output/annonars/gnomad-exomes-grch38-{DV.gnomad_v2}+{PV.annonars}/rocksdb/IDENTITY",
+        f"output/annonars/gnomad-genomes-grch37-{DV.gnomad_v2}+{PV.annonars}/rocksdb/IDENTITY",
+        f"output/annonars/gnomad-genomes-grch38-{DV.gnomad_v3}+{PV.annonars}/rocksdb/IDENTITY",
+        f"output/annonars/helixmtdb-grch37-{DV.helixmtdb}+{PV.annonars}/rocksdb/IDENTITY",
+        f"output/annonars/helixmtdb-grch38-{DV.helixmtdb}+{PV.annonars}/rocksdb/IDENTITY",
+        f"output/annonars/cons-grch37-{DV.ucsc_cons_37}+{PV.annonars}/rocksdb/IDENTITY",
+        f"output/annonars/cons-grch38-{DV.ucsc_cons_38}+{PV.annonars}/rocksdb/IDENTITY",
         # ----- Genes
         f"output/worker/genes-{DV.acmg_sf}+{DV.gnomad_constraints}+{DV.dbnsfp}+{DV.today}+{PV.worker}/rocksdb/IDENTITY",
         f"output/worker/genes-xlink-{DV.today}/genes-xlink.tsv",
diff --git a/rules/output/annonars/cadd.smk b/rules/output/annonars/cadd.smk
@@ -49,7 +49,7 @@ rule output_annonars_cadd:  # -- build CADD RocksDB with annonars
     input:
         unpack(input_output_annonars_cadd),
     output:
-        "output/annonars/annos/seqvars/cadd-{genome_release}-{v_cadd}+{v_annonars}/rocksdb/IDENTITY",
+        "output/annonars/cadd-{genome_release}-{v_cadd}+{v_annonars}/rocksdb/IDENTITY",
     threads: int(os.environ.get("THREADS_ANNONARS_IMPORT", "96"))
     resources:
         runtime=os.environ.get("RUNTIME_ANNONARS_IMPORT", "48h"),
diff --git a/rules/output/annonars/cons.smk b/rules/output/annonars/cons.smk
@@ -7,7 +7,7 @@ rule output_annonars_cons:  # -- build UCSC conservation track RocksDB with anno
     input:
         tsv="work/annos/{genome_release}/features/cons/{v_cons}/ucsc_conservation.tsv",
     output:
-        "output/annonars/annos/seqvars/cons-{genome_release}-{v_cons}+{v_annonars}/rocksdb/IDENTITY",
+        "output/annonars/cons-{genome_release}-{v_cons}+{v_annonars}/rocksdb/IDENTITY",
     threads: int(os.environ.get("THREADS_ANNONARS_IMPORT", "96"))
     resources:
         runtime=os.environ.get("RUNTIME_ANNONARS_IMPORT", "48h"),
diff --git a/rules/output/annonars/dbnsfp.smk b/rules/output/annonars/dbnsfp.smk
@@ -12,7 +12,7 @@ rule output_annonars_dbnsfp:  # -- build dbNSFP RocksDB with annonars
     input:
         input_output_annonars_dbnsfp,
     output:
-        "output/annonars/annos/seqvars/dbnsfp-{genome_release}-{v_dbnsfp}+{v_annonars}/rocksdb/IDENTITY",
+        "output/annonars/dbnsfp-{genome_release}-{v_dbnsfp}+{v_annonars}/rocksdb/IDENTITY",
     threads: int(os.environ.get("THREADS_ANNONARS_IMPORT", "96"))
     resources:
         runtime=os.environ.get("RUNTIME_ANNONARS_IMPORT", "48h"),
diff --git a/rules/output/annonars/dbscsnv.smk b/rules/output/annonars/dbscsnv.smk
@@ -12,7 +12,7 @@ rule output_annonars_dbscsnv:  # -- build dbscSNV RocksDB with annonars
     input:
         input_output_annonars_dbscsnv,
     output:
-        "output/annonars/annos/seqvars/dbscsnv-{genome_release}-{v_dbscsnv}+{v_annonars}/rocksdb/IDENTITY",
+        "output/annonars/dbscsnv-{genome_release}-{v_dbscsnv}+{v_annonars}/rocksdb/IDENTITY",
     threads: int(os.environ.get("THREADS_ANNONARS_IMPORT", "96"))
     resources:
         runtime=os.environ.get("RUNTIME_ANNONARS_IMPORT", "48h"),
diff --git a/rules/output/annonars/dbsnp.smk b/rules/output/annonars/dbsnp.smk
@@ -7,7 +7,7 @@ rule output_annonars_dbsnp:  # -- build dbSNP RocksDB with annonars
     input:
         vcf="work/download/annos/{genome_release}/seqvars/dbsnp/{v_dbsnp}/dbsnp.vcf.gz",
     output:
-        "output/annonars/annos/seqvars/dbsnp-{genome_release}-{v_dbsnp}+{v_annonars}/rocksdb/IDENTITY",
+        "output/annonars/dbsnp-{genome_release}-{v_dbsnp}+{v_annonars}/rocksdb/IDENTITY",
     threads: int(os.environ.get("THREADS_ANNONARS_IMPORT", "96"))
     resources:
         runtime=os.environ.get("RUNTIME_ANNONARS_IMPORT", "48h"),
diff --git a/rules/output/annonars/gnomad_exomes.smk b/rules/output/annonars/gnomad_exomes.smk
@@ -7,7 +7,7 @@ rule output_annonars_gnomad_exomes:  # -- build gnomAD-exomes RocksDB with annon
     input:
         vcf="work/download/annos/{genome_release}/seqvars/gnomad_exomes/{v_gnomad}/.done",
     output:
-        "output/annonars/annos/seqvars/gnomad-exomes-{genome_release}-{v_gnomad}+{v_annonars}/rocksdb/IDENTITY",
+        "output/annonars/gnomad-exomes-{genome_release}-{v_gnomad}+{v_annonars}/rocksdb/IDENTITY",
     threads: int(os.environ.get("THREADS_ANNONARS_IMPORT", "96"))
     resources:
         runtime=os.environ.get("RUNTIME_ANNONARS_IMPORT", "48h"),
diff --git a/rules/output/annonars/gnomad_genomes.smk b/rules/output/annonars/gnomad_genomes.smk
@@ -7,7 +7,7 @@ rule output_annonars_gnomad_genomes:  # -- build gnomAD-genomes RocksDB with ann
     input:
         vcf="work/download/annos/{genome_release}/seqvars/gnomad_genomes/{v_gnomad}/.done",
     output:
-        "output/annonars/annos/seqvars/gnomad-genomes-{genome_release}-{v_gnomad}+{v_annonars}/rocksdb/IDENTITY",
+        "output/annonars/gnomad-genomes-{genome_release}-{v_gnomad}+{v_annonars}/rocksdb/IDENTITY",
     threads: int(os.environ.get("THREADS_ANNONARS_IMPORT", "96"))
     resources:
         runtime=os.environ.get("RUNTIME_ANNONARS_IMPORT", "48h"),
diff --git a/rules/output/annonars/gnomad_mtdna.smk b/rules/output/annonars/gnomad_mtdna.smk
@@ -7,7 +7,7 @@ rule output_annonars_gnomad_mtdna:  # -- build gnomAD-mtDNA RocksDB with annonar
     input:
         vcf="work/annos/{genome_release}/seqvars/gnomad_mtdna/{v_gnomad}/gnomad_mtdna.vcf.gz",
     output:
-        "output/annonars/annos/seqvars/gnomad-mtdna-{genome_release}-{v_gnomad}+{v_annonars}/rocksdb/IDENTITY",
+        "output/annonars/gnomad-mtdna-{genome_release}-{v_gnomad}+{v_annonars}/rocksdb/IDENTITY",
     threads: int(os.environ.get("THREADS_ANNONARS_IMPORT", "96"))
     resources:
         runtime=os.environ.get("RUNTIME_ANNONARS_IMPORT", "48h"),
diff --git a/rules/output/annonars/helix.smk b/rules/output/annonars/helix.smk
@@ -7,7 +7,7 @@ rule output_annonars_helixmtdb:  # -- build HelixMtDb RocksDB with annonars
     input:
         vcf="work/annos/{genome_release}/seqvars/helixmtdb/{v_helixmtdb}/helixmtdb.vcf.gz",
     output:
-        "output/annonars/annos/seqvars/helixmtdb-{genome_release}-{v_helixmtdb}+{v_annonars}/rocksdb/IDENTITY",
+        "output/annonars/helixmtdb-{genome_release}-{v_helixmtdb}+{v_annonars}/rocksdb/IDENTITY",
     threads: int(os.environ.get("THREADS_ANNONARS_IMPORT", "96"))
     resources:
         runtime=os.environ.get("RUNTIME_ANNONARS_IMPORT", "48h"),
diff --git a/rules/output/viguno/hpo.smk b/rules/output/viguno/hpo.smk
@@ -16,7 +16,7 @@ rule output_worker_pheno:  # -- copy HPO and simulate
         obo="output/viguno/hpo-{v_hpo}+{v_viguno}/hp.obo",
         hpoa="output/viguno/hpo-{v_hpo}+{v_viguno}/phenotype.hpoa",
         phenotype_to_genes="output/viguno/hpo-{v_hpo}+{v_viguno}/phenotype_to_genes.txt",
-        rocksdb_identity="output/viguno/hpo-{v_hpo}+{v_viguno}/resnik/IDENTITY",
+        rocksdb_identity="output/viguno/hpo-{v_hpo}+{v_viguno}/scores-fun-sim-avg-resnik-gene/IDENTITY",
     wildcard_constraints:
         v_hpo=RE_VERSION,
         v_viguno=RE_VERSION,
@@ -34,10 +34,10 @@ rule output_worker_pheno:  # -- copy HPO and simulate
 
         viguno simulate \
             --ic-base gene \
-            --similarity fun-sim-avg \
-            --combiner resnik \
+            --similarity resnik \
+            --combiner fun-sim-avg \
             --path-hpo-dir $(dirname {input.obo}) \
-            --path-out-rocksdb $(dirname {output.rocksdb_identity})/scores-fun-sim-avg-resnik-gene \
+            --path-out-rocksdb $(dirname {output.rocksdb_identity}) \
             --min-terms 1 \
             $(if [[ "{RUNS_IN_CI}" == "True" ]]; then \
                 echo --max-terms 1; \
@@ -52,12 +52,12 @@ rule output_worker_pheno:  # -- copy HPO and simulate
 
 rule global_hpo_to_bin:  # -- convert to .bin
     input:
-        obo="output/viguno/hpo-{v_hpo}+{v_viguno}/hp.obo",
-        hpoa="output/viguno/hpo-{v_hpo}+{v_viguno}/phenotype.hpoa",
-        phenotype_to_genes="output/viguno/hpo-{v_hpo}+{v_viguno}/phenotype_to_genes.txt",
-        rocksdb_identity="output/viguno/hpo-{v_hpo}+{v_viguno}/resnik/IDENTITY",
+        obo="work/download/hpo/{v_hpo}/hp.obo",
+        hpoa="work/download/hpo/{v_hpo}/phenotype.hpoa",
+        genes_to_phenotype="work/download/hpo/{v_hpo}/phenotype_to_genes.txt",
+        rocksdb_identity="output/viguno/hpo-{v_hpo}+{v_viguno}/scores-fun-sim-avg-resnik-gene/IDENTITY",
     output:
-        obo="output/viguno/hpo-{v_hpo}+{v_viguno}/hpo.bin",
+        bin="output/viguno/hpo-{v_hpo}+{v_viguno}/hpo.bin",
     wildcard_constraints:
         v_hpo=RE_VERSION,
         v_viguno=RE_VERSION,
