feat: provide enrichment probesets for Agilent/IDT/Twist (#52) (#53)

varfish-org · Jun 29, 2023 · 618941c · 618941c
1 parent ba84443
commit 618941c
Show file tree

Hide file tree

Showing 47 changed files with 174 additions and 0 deletions.
diff --git a/.gitattributes b/.gitattributes
@@ -1 +1,2 @@
 excerpt-data/**/* filter=lfs diff=lfs merge=lfs -text
+bundled-data/**/* filter=lfs diff=lfs merge=lfs -binary
diff --git a/Snakefile b/Snakefile
@@ -218,6 +218,29 @@ rule all:
         # ----- tads
         f"output/full/tracks/track-tads-grch37-dixon2015+{DV.tracks}/hesc.bed",
         f"output/full/tracks/track-tads-grch38-dixon2015+{DV.tracks}/hesc.bed",
+        # ----- probesets
+        f"output/full/tracks/track-enrichment-probesets-targets/agilent-all-exon-v4-grch37-{DV.tracks}.bed.gz",
+        f"output/full/tracks/track-enrichment-probesets-targets/agilent-all-exon-v4-grch38-{DV.tracks}.bed.gz",
+        f"output/full/tracks/track-enrichment-probesets-targets/agilent-all-exon-v5-grch37-{DV.tracks}.bed.gz",
+        f"output/full/tracks/track-enrichment-probesets-targets/agilent-all-exon-v5-grch38-{DV.tracks}.bed.gz",
+        f"output/full/tracks/track-enrichment-probesets-targets/agilent-all-exon-v6-grch37-{DV.tracks}.bed.gz",
+        f"output/full/tracks/track-enrichment-probesets-targets/agilent-all-exon-v6-grch38-{DV.tracks}.bed.gz",
+        f"output/full/tracks/track-enrichment-probesets-targets/agilent-all-exon-v7-grch37-{DV.tracks}.bed.gz",
+        f"output/full/tracks/track-enrichment-probesets-targets/agilent-all-exon-v7-grch38-{DV.tracks}.bed.gz",
+        f"output/full/tracks/track-enrichment-probesets-targets/agilent-all-exon-v8-grch37-{DV.tracks}.bed.gz",
+        f"output/full/tracks/track-enrichment-probesets-targets/agilent-all-exon-v8-grch38-{DV.tracks}.bed.gz",
+        f"output/full/tracks/track-enrichment-probesets-targets/idt-xgen-exome-research-panel-v1-grch37-{DV.tracks}.bed.gz",
+        f"output/full/tracks/track-enrichment-probesets-targets/idt-xgen-exome-research-panel-v1-grch38-{DV.tracks}.bed.gz",
+        f"output/full/tracks/track-enrichment-probesets-targets/idt-xgen-exome-research-panel-v2-grch37-{DV.tracks}.bed.gz",
+        f"output/full/tracks/track-enrichment-probesets-targets/idt-xgen-exome-research-panel-v2-grch38-{DV.tracks}.bed.gz",
+        f"output/full/tracks/track-enrichment-probesets-targets/twist-comprehensive-exome-grch37-{DV.tracks}.bed.gz",
+        f"output/full/tracks/track-enrichment-probesets-targets/twist-comprehensive-exome-grch38-{DV.tracks}.bed.gz",
+        f"output/full/tracks/track-enrichment-probesets-targets/twist-core-exome-grch37-{DV.tracks}.bed.gz",
+        f"output/full/tracks/track-enrichment-probesets-targets/twist-core-exome-grch38-{DV.tracks}.bed.gz",
+        f"output/full/tracks/track-enrichment-probesets-targets/twist-exome-v2_0-grch37-{DV.tracks}.bed.gz",
+        f"output/full/tracks/track-enrichment-probesets-targets/twist-exome-v2_0-grch38-{DV.tracks}.bed.gz",
+        f"output/full/tracks/track-enrichment-probesets-targets/twist-refseq-exome-grch37-{DV.tracks}.bed.gz",
+        f"output/full/tracks/track-enrichment-probesets-targets/twist-refseq-exome-grch38-{DV.tracks}.bed.gz",
         #
         # == development (reduced data) directories =============================================
         #
@@ -348,6 +371,8 @@ include: "rules/output/worker/mim2gene.smk"
 include: "rules/output/worker/masked.smk"
 include: "rules/output/worker/bgdb.smk"
 include: "rules/output/worker/tads.smk"
+# ---- tracks
+include: "rules/output/tracks/exome_probesets.smk"
 # -- reduced output directory (dev/exomes) ------------------------------------------------------
 # ---- bed file
 include: "rules/reduced/annonars.smk"

diff --git a/bundled-data/track-enrichment-probesets/agilent-all-exon-v4-targets-grch37.bed.gz b/bundled-data/track-enrichment-probesets/agilent-all-exon-v4-targets-grch37.bed.gz
diff --git a/bundled-data/track-enrichment-probesets/agilent-all-exon-v4-targets-grch37.bed.gz.tbi b/bundled-data/track-enrichment-probesets/agilent-all-exon-v4-targets-grch37.bed.gz.tbi
diff --git a/bundled-data/track-enrichment-probesets/agilent-all-exon-v4-targets-grch38.bed.gz b/bundled-data/track-enrichment-probesets/agilent-all-exon-v4-targets-grch38.bed.gz
diff --git a/bundled-data/track-enrichment-probesets/agilent-all-exon-v4-targets-grch38.bed.gz.tbi b/bundled-data/track-enrichment-probesets/agilent-all-exon-v4-targets-grch38.bed.gz.tbi
diff --git a/bundled-data/track-enrichment-probesets/agilent-all-exon-v5-targets-grch37.bed.gz b/bundled-data/track-enrichment-probesets/agilent-all-exon-v5-targets-grch37.bed.gz
diff --git a/bundled-data/track-enrichment-probesets/agilent-all-exon-v5-targets-grch37.bed.gz.tbi b/bundled-data/track-enrichment-probesets/agilent-all-exon-v5-targets-grch37.bed.gz.tbi
diff --git a/bundled-data/track-enrichment-probesets/agilent-all-exon-v5-targets-grch38.bed.gz b/bundled-data/track-enrichment-probesets/agilent-all-exon-v5-targets-grch38.bed.gz
diff --git a/bundled-data/track-enrichment-probesets/agilent-all-exon-v5-targets-grch38.bed.gz.tbi b/bundled-data/track-enrichment-probesets/agilent-all-exon-v5-targets-grch38.bed.gz.tbi
diff --git a/bundled-data/track-enrichment-probesets/agilent-all-exon-v6-targets-grch37.bed.gz b/bundled-data/track-enrichment-probesets/agilent-all-exon-v6-targets-grch37.bed.gz
diff --git a/bundled-data/track-enrichment-probesets/agilent-all-exon-v6-targets-grch37.bed.gz.tbi b/bundled-data/track-enrichment-probesets/agilent-all-exon-v6-targets-grch37.bed.gz.tbi
diff --git a/bundled-data/track-enrichment-probesets/agilent-all-exon-v6-targets-grch38.bed.gz b/bundled-data/track-enrichment-probesets/agilent-all-exon-v6-targets-grch38.bed.gz
diff --git a/bundled-data/track-enrichment-probesets/agilent-all-exon-v6-targets-grch38.bed.gz.tbi b/bundled-data/track-enrichment-probesets/agilent-all-exon-v6-targets-grch38.bed.gz.tbi
diff --git a/bundled-data/track-enrichment-probesets/agilent-all-exon-v7-targets-grch37.bed.gz b/bundled-data/track-enrichment-probesets/agilent-all-exon-v7-targets-grch37.bed.gz
diff --git a/bundled-data/track-enrichment-probesets/agilent-all-exon-v7-targets-grch37.bed.gz.tbi b/bundled-data/track-enrichment-probesets/agilent-all-exon-v7-targets-grch37.bed.gz.tbi
diff --git a/bundled-data/track-enrichment-probesets/agilent-all-exon-v7-targets-grch38.bed.gz b/bundled-data/track-enrichment-probesets/agilent-all-exon-v7-targets-grch38.bed.gz
diff --git a/bundled-data/track-enrichment-probesets/agilent-all-exon-v7-targets-grch38.bed.gz.tbi b/bundled-data/track-enrichment-probesets/agilent-all-exon-v7-targets-grch38.bed.gz.tbi
diff --git a/bundled-data/track-enrichment-probesets/agilent-all-exon-v8-targets-grch37.bed.gz b/bundled-data/track-enrichment-probesets/agilent-all-exon-v8-targets-grch37.bed.gz
diff --git a/bundled-data/track-enrichment-probesets/agilent-all-exon-v8-targets-grch37.bed.gz.tbi b/bundled-data/track-enrichment-probesets/agilent-all-exon-v8-targets-grch37.bed.gz.tbi
diff --git a/bundled-data/track-enrichment-probesets/agilent-all-exon-v8-targets-grch38.bed.gz b/bundled-data/track-enrichment-probesets/agilent-all-exon-v8-targets-grch38.bed.gz
diff --git a/bundled-data/track-enrichment-probesets/agilent-all-exon-v8-targets-grch38.bed.gz.tbi b/bundled-data/track-enrichment-probesets/agilent-all-exon-v8-targets-grch38.bed.gz.tbi
diff --git a/...ed-data/track-enrichment-probesets/idt-xgen-exome-research-panel-v1-targets-grch37.bed.gz b/...ed-data/track-enrichment-probesets/idt-xgen-exome-research-panel-v1-targets-grch37.bed.gz
diff --git a/...ata/track-enrichment-probesets/idt-xgen-exome-research-panel-v1-targets-grch37.bed.gz.tbi b/...ata/track-enrichment-probesets/idt-xgen-exome-research-panel-v1-targets-grch37.bed.gz.tbi
diff --git a/...ed-data/track-enrichment-probesets/idt-xgen-exome-research-panel-v1-targets-grch38.bed.gz b/...ed-data/track-enrichment-probesets/idt-xgen-exome-research-panel-v1-targets-grch38.bed.gz
diff --git a/...ata/track-enrichment-probesets/idt-xgen-exome-research-panel-v1-targets-grch38.bed.gz.tbi b/...ata/track-enrichment-probesets/idt-xgen-exome-research-panel-v1-targets-grch38.bed.gz.tbi
diff --git a/...ed-data/track-enrichment-probesets/idt-xgen-exome-research-panel-v2-targets-grch37.bed.gz b/...ed-data/track-enrichment-probesets/idt-xgen-exome-research-panel-v2-targets-grch37.bed.gz
diff --git a/...ata/track-enrichment-probesets/idt-xgen-exome-research-panel-v2-targets-grch37.bed.gz.tbi b/...ata/track-enrichment-probesets/idt-xgen-exome-research-panel-v2-targets-grch37.bed.gz.tbi
diff --git a/...ed-data/track-enrichment-probesets/idt-xgen-exome-research-panel-v2-targets-grch38.bed.gz b/...ed-data/track-enrichment-probesets/idt-xgen-exome-research-panel-v2-targets-grch38.bed.gz
diff --git a/...ata/track-enrichment-probesets/idt-xgen-exome-research-panel-v2-targets-grch38.bed.gz.tbi b/...ata/track-enrichment-probesets/idt-xgen-exome-research-panel-v2-targets-grch38.bed.gz.tbi
diff --git a/bundled-data/track-enrichment-probesets/twist-comprehensive-exome-targets-grch37.bed.gz b/bundled-data/track-enrichment-probesets/twist-comprehensive-exome-targets-grch37.bed.gz
diff --git a/bundled-data/track-enrichment-probesets/twist-comprehensive-exome-targets-grch37.bed.gz.tbi b/bundled-data/track-enrichment-probesets/twist-comprehensive-exome-targets-grch37.bed.gz.tbi
diff --git a/bundled-data/track-enrichment-probesets/twist-comprehensive-exome-targets-grch38.bed.gz b/bundled-data/track-enrichment-probesets/twist-comprehensive-exome-targets-grch38.bed.gz
diff --git a/bundled-data/track-enrichment-probesets/twist-comprehensive-exome-targets-grch38.bed.gz.tbi b/bundled-data/track-enrichment-probesets/twist-comprehensive-exome-targets-grch38.bed.gz.tbi
diff --git a/bundled-data/track-enrichment-probesets/twist-core-exome-targets-grch37.bed.gz b/bundled-data/track-enrichment-probesets/twist-core-exome-targets-grch37.bed.gz
diff --git a/bundled-data/track-enrichment-probesets/twist-core-exome-targets-grch37.bed.gz.tbi b/bundled-data/track-enrichment-probesets/twist-core-exome-targets-grch37.bed.gz.tbi
diff --git a/bundled-data/track-enrichment-probesets/twist-core-exome-targets-grch38.bed.gz b/bundled-data/track-enrichment-probesets/twist-core-exome-targets-grch38.bed.gz
diff --git a/bundled-data/track-enrichment-probesets/twist-core-exome-targets-grch38.bed.gz.tbi b/bundled-data/track-enrichment-probesets/twist-core-exome-targets-grch38.bed.gz.tbi
diff --git a/bundled-data/track-enrichment-probesets/twist-exome-v2_0-targets-grch37.bed.gz b/bundled-data/track-enrichment-probesets/twist-exome-v2_0-targets-grch37.bed.gz
diff --git a/bundled-data/track-enrichment-probesets/twist-exome-v2_0-targets-grch37.bed.gz.tbi b/bundled-data/track-enrichment-probesets/twist-exome-v2_0-targets-grch37.bed.gz.tbi
diff --git a/bundled-data/track-enrichment-probesets/twist-exome-v2_0-targets-grch38.bed.gz b/bundled-data/track-enrichment-probesets/twist-exome-v2_0-targets-grch38.bed.gz
diff --git a/bundled-data/track-enrichment-probesets/twist-exome-v2_0-targets-grch38.bed.gz.tbi b/bundled-data/track-enrichment-probesets/twist-exome-v2_0-targets-grch38.bed.gz.tbi
diff --git a/bundled-data/track-enrichment-probesets/twist-refseq-exome-targets-grch37.bed.gz b/bundled-data/track-enrichment-probesets/twist-refseq-exome-targets-grch37.bed.gz
diff --git a/bundled-data/track-enrichment-probesets/twist-refseq-exome-targets-grch37.bed.gz.tbi b/bundled-data/track-enrichment-probesets/twist-refseq-exome-targets-grch37.bed.gz.tbi
diff --git a/bundled-data/track-enrichment-probesets/twist-refseq-exome-targets-grch38.bed.gz b/bundled-data/track-enrichment-probesets/twist-refseq-exome-targets-grch38.bed.gz
diff --git a/bundled-data/track-enrichment-probesets/twist-refseq-exome-targets-grch38.bed.gz.tbi b/bundled-data/track-enrichment-probesets/twist-refseq-exome-targets-grch38.bed.gz.tbi
diff --git a/rules/output/tracks/exome_probesets.smk b/rules/output/tracks/exome_probesets.smk
@@ -0,0 +1,16 @@
+## Rules related to exome probesets.
+
+
+rule output_tracks_exome_probeset:
+    output:
+        bed="output/full/tracks/track-enrichment-probesets-targets/{name}-{genome_release}-{v_tracks}.bed.gz",
+        bed_tbi="output/full/tracks/track-enrichment-probesets-targets/{name}-{genome_release}-{v_tracks}.bed.gz.tbi",
+    wildcard_constraints:
+        name=r"[a-zA-Z0-9_-]+",
+        genome_release=RE_GENOME,
+        v_tracks=RE_VERSION,
+    shell:
+        r"""
+        cp bundled-data/track-enrichment-probesetss/{wildcards.name}-targets-{wildcards.genome_release}.bed.gz {output.bed}
+        cp bundled-data/track-enrichment-probesetss/{wildcards.name}-targets-{wildcards.genome_release}.bed.gz.tbi {output.bed_tbi}
+        """
Original file line number	Diff line number	Diff line change
		@@ -1 +1,2 @@
		excerpt-data/*/ filter=lfs diff=lfs merge=lfs -text
		bundled-data/*/ filter=lfs diff=lfs merge=lfs -binary