diff --git a/Snakefile b/Snakefile index fe6da1e..22b880d 100644 --- a/Snakefile +++ b/Snakefile @@ -127,7 +127,6 @@ rule all: f"work/annos/grch38/seqvars/gnomad_mtdna/{DV.gnomad_mtdna}/gnomad_mtdna.vcf.gz", f"work/download/annos/grch38/seqvars/gnomad_exomes/{DV.gnomad_v4}/.done", f"work/download/annos/grch38/seqvars/gnomad_genomes/{DV.gnomad_v4}/.done", - # NB: gnomAD-SV GRCh38 was announced end of 2020 but not released yet # -- genome browser "features" (position-specific) # ---- GRCh37 f"work/annos/grch37/features/cons/{DV.ucsc_cons_37}/ucsc_conservation.tsv", @@ -200,6 +199,8 @@ rule all: f"output/full/worker/bgdb-dgv-gs-grch38-{DV.dgv}+{PV.worker}/bgdb-dgv-gs.bin", f"output/full/worker/bgdb-gnomad-grch37-{DV.gnomad_sv}+{PV.worker}/bgdb-gnomad.bin", f"output/full/worker/bgdb-exac-grch37-{DV.exac_cnv}+{PV.worker}/bgdb-exac.bin", + f"output/full/worker/bgdb-gnomad-exomes-cnv-grch38-{DV.gnomad_sv4}+{PV.worker}/bgdb-gnomad-exomes-cnv-grch38.bin", + f"output/full/worker/bgdb-gnomad-genomes-sv-grch38-{DV.gnomad_sv4}+{PV.worker}/bgdb-gnomad-genomes-sv-grch38.bin", f"output/full/worker/bgdb-g1k-grch37-{DV.g1k_svs}+{PV.worker}/bgdb-g1k.bin", f"output/full/worker/clinvar-strucvars-grch37-{DV.clinvar_version}+{PV.worker}/clinvar-strucvars.bin", f"output/full/worker/clinvar-strucvars-grch38-{DV.clinvar_version}+{PV.worker}/clinvar-strucvars.bin", @@ -227,6 +228,8 @@ rule all: f"output/full/tracks/track-strucvars-exac-grch37-{DV.exac_cnv}+{DV.tracks}/exac.bed.gz", f"output/full/tracks/track-strucvars-g1k-grch37-{DV.g1k_svs}+{DV.tracks}/g1k.bed.gz", f"output/full/tracks/track-strucvars-gnomad-grch37-{DV.gnomad_sv}+{DV.tracks}/gnomad.bed.gz", + f"output/full/tracks/track-strucvars-gnomad-sv-grch38-{DV.gnomad_sv4}+{DV.tracks}/gnomad-sv.bed.gz", + f"output/full/tracks/track-strucvars-gnomad-cnv-grch38-{DV.gnomad_cnv4}+{DV.tracks}/gnomad-cnv.bed.gz", # ----- known pathogenic MMS f"output/full/tracks/track-strucvars-patho-mms-grch37-{DV.patho_mms}+{DV.tracks}/patho-mms.bed", f"output/full/tracks/track-strucvars-patho-mms-grch38-{DV.patho_mms}+{DV.tracks}/patho-mms.bed", diff --git a/download_urls.yml b/download_urls.yml index 4b3ecea..3e3cf55 100644 --- a/download_urls.yml +++ b/download_urls.yml @@ -105,29 +105,101 @@ strategy: gz-head count: 160 - url: https://storage.googleapis.com/gcp-public-data--gnomad/release/4.0/genome_sv/gnomad.v4.0.sv.chr1.vcf.gz + excerpt_strategy: + strategy: gz-head + count: 1800 - url: https://storage.googleapis.com/gcp-public-data--gnomad/release/4.0/genome_sv/gnomad.v4.0.sv.chr2.vcf.gz + excerpt_strategy: + strategy: gz-head + count: 1800 - url: https://storage.googleapis.com/gcp-public-data--gnomad/release/4.0/genome_sv/gnomad.v4.0.sv.chr3.vcf.gz + excerpt_strategy: + strategy: gz-head + count: 1800 - url: https://storage.googleapis.com/gcp-public-data--gnomad/release/4.0/genome_sv/gnomad.v4.0.sv.chr4.vcf.gz + excerpt_strategy: + strategy: gz-head + count: 1800 - url: https://storage.googleapis.com/gcp-public-data--gnomad/release/4.0/genome_sv/gnomad.v4.0.sv.chr5.vcf.gz + excerpt_strategy: + strategy: gz-head + count: 1800 - url: https://storage.googleapis.com/gcp-public-data--gnomad/release/4.0/genome_sv/gnomad.v4.0.sv.chr6.vcf.gz + excerpt_strategy: + strategy: gz-head + count: 1800 - url: https://storage.googleapis.com/gcp-public-data--gnomad/release/4.0/genome_sv/gnomad.v4.0.sv.chr7.vcf.gz + excerpt_strategy: + strategy: gz-head + count: 1800 - url: https://storage.googleapis.com/gcp-public-data--gnomad/release/4.0/genome_sv/gnomad.v4.0.sv.chr8.vcf.gz + excerpt_strategy: + strategy: gz-head + count: 1800 - url: https://storage.googleapis.com/gcp-public-data--gnomad/release/4.0/genome_sv/gnomad.v4.0.sv.chr9.vcf.gz + excerpt_strategy: + strategy: gz-head + count: 1800 - url: https://storage.googleapis.com/gcp-public-data--gnomad/release/4.0/genome_sv/gnomad.v4.0.sv.chr10.vcf.gz + excerpt_strategy: + strategy: gz-head + count: 1800 - url: https://storage.googleapis.com/gcp-public-data--gnomad/release/4.0/genome_sv/gnomad.v4.0.sv.chr11.vcf.gz + excerpt_strategy: + strategy: gz-head + count: 1800 - url: https://storage.googleapis.com/gcp-public-data--gnomad/release/4.0/genome_sv/gnomad.v4.0.sv.chr12.vcf.gz + excerpt_strategy: + strategy: gz-head + count: 1800 - url: https://storage.googleapis.com/gcp-public-data--gnomad/release/4.0/genome_sv/gnomad.v4.0.sv.chr13.vcf.gz + excerpt_strategy: + strategy: gz-head + count: 1800 - url: https://storage.googleapis.com/gcp-public-data--gnomad/release/4.0/genome_sv/gnomad.v4.0.sv.chr14.vcf.gz + excerpt_strategy: + strategy: gz-head + count: 1800 - url: https://storage.googleapis.com/gcp-public-data--gnomad/release/4.0/genome_sv/gnomad.v4.0.sv.chr15.vcf.gz + excerpt_strategy: + strategy: gz-head + count: 1800 - url: https://storage.googleapis.com/gcp-public-data--gnomad/release/4.0/genome_sv/gnomad.v4.0.sv.chr16.vcf.gz + excerpt_strategy: + strategy: gz-head + count: 1800 - url: https://storage.googleapis.com/gcp-public-data--gnomad/release/4.0/genome_sv/gnomad.v4.0.sv.chr17.vcf.gz + excerpt_strategy: + strategy: gz-head + count: 1800 - url: https://storage.googleapis.com/gcp-public-data--gnomad/release/4.0/genome_sv/gnomad.v4.0.sv.chr18.vcf.gz + excerpt_strategy: + strategy: gz-head + count: 1800 - url: https://storage.googleapis.com/gcp-public-data--gnomad/release/4.0/genome_sv/gnomad.v4.0.sv.chr19.vcf.gz + excerpt_strategy: + strategy: gz-head + count: 1800 - url: https://storage.googleapis.com/gcp-public-data--gnomad/release/4.0/genome_sv/gnomad.v4.0.sv.chr20.vcf.gz + excerpt_strategy: + strategy: gz-head + count: 1800 - url: https://storage.googleapis.com/gcp-public-data--gnomad/release/4.0/genome_sv/gnomad.v4.0.sv.chr21.vcf.gz + excerpt_strategy: + strategy: gz-head + count: 1800 - url: https://storage.googleapis.com/gcp-public-data--gnomad/release/4.0/genome_sv/gnomad.v4.0.sv.chr22.vcf.gz + excerpt_strategy: + strategy: gz-head + count: 1800 - url: https://storage.googleapis.com/gcp-public-data--gnomad/release/4.0/genome_sv/gnomad.v4.0.sv.chrX.vcf.gz + excerpt_strategy: + strategy: gz-head + count: 1800 - url: https://storage.googleapis.com/gcp-public-data--gnomad/release/4.0/genome_sv/gnomad.v4.0.sv.chrY.vcf.gz + excerpt_strategy: + strategy: gz-head + count: 1800 - url: https://www.deciphergenomics.org/files/downloads/HI_Predictions_Version3.bed.gz diff --git a/environment.yml b/environment.yml index 5e552bb..fc74e6d 100644 --- a/environment.yml +++ b/environment.yml @@ -46,7 +46,7 @@ dependencies: - annonars =0.34.0 - viguno =0.2.0 - mehari =0.21.1 - - varfish-server-worker =0.10.2 + - varfish-server-worker =0.12.0 # S3 uploads - s5cmd =2.1.0 # async HTTP requests diff --git a/excerpt-data/05e93e6f1f5d60e6/gnomad.v4.0.sv.chr4.vcf.gz b/excerpt-data/05e93e6f1f5d60e6/gnomad.v4.0.sv.chr4.vcf.gz index fb9a00e..9e65e06 100644 --- a/excerpt-data/05e93e6f1f5d60e6/gnomad.v4.0.sv.chr4.vcf.gz +++ b/excerpt-data/05e93e6f1f5d60e6/gnomad.v4.0.sv.chr4.vcf.gz @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:74cdc80bc74dd499a52d06dea268987036ea52bdab9d9add9c5e3816a4629f43 -size 420152 +oid sha256:4e28fd287c75f30a72a03849afbb8a325dd25e2558964ee557e3b01397a8b065 +size 6164 diff --git a/excerpt-data/05e93e6f1f5d60e6/gnomad.v4.0.sv.chr4.vcf.gz.tbi b/excerpt-data/05e93e6f1f5d60e6/gnomad.v4.0.sv.chr4.vcf.gz.tbi deleted file mode 100644 index ad46be9..0000000 --- a/excerpt-data/05e93e6f1f5d60e6/gnomad.v4.0.sv.chr4.vcf.gz.tbi +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:952a07ed29b3360170a993dc12a1aa6773820233779762704655446b32baa416 -size 414 diff --git a/excerpt-data/0be9b2561c9397f2/gnomad.v4.0.sv.chr15.vcf.gz b/excerpt-data/0be9b2561c9397f2/gnomad.v4.0.sv.chr15.vcf.gz index 6653e09..61382e7 100644 --- a/excerpt-data/0be9b2561c9397f2/gnomad.v4.0.sv.chr15.vcf.gz +++ b/excerpt-data/0be9b2561c9397f2/gnomad.v4.0.sv.chr15.vcf.gz @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:d891ea2a13db923a679f0f38578a4ed30518bdb9ea41b21585e9250d8d5c19e7 -size 489677 +oid sha256:c33337889d635bf0ade57d2ffa559e7e00f040e08c3e236f0e9fc05002dc13c2 +size 6159 diff --git a/excerpt-data/0be9b2561c9397f2/gnomad.v4.0.sv.chr15.vcf.gz.tbi b/excerpt-data/0be9b2561c9397f2/gnomad.v4.0.sv.chr15.vcf.gz.tbi deleted file mode 100644 index ae0fcdb..0000000 --- a/excerpt-data/0be9b2561c9397f2/gnomad.v4.0.sv.chr15.vcf.gz.tbi +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:6ccc9925e01347b6290c9a7d57bb3d1ece3a68eac2c44cf840dcddf84c0b2307 -size 296 diff --git a/excerpt-data/13d4b1406e769b80/gnomad.v4.0.sv.chr3.vcf.gz b/excerpt-data/13d4b1406e769b80/gnomad.v4.0.sv.chr3.vcf.gz index aac91a5..a320a36 100644 --- a/excerpt-data/13d4b1406e769b80/gnomad.v4.0.sv.chr3.vcf.gz +++ b/excerpt-data/13d4b1406e769b80/gnomad.v4.0.sv.chr3.vcf.gz @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:cae509c9c8a04c8be9606de238b74e158873bc1e0a22480701a0dafa40df2849 -size 454927 +oid sha256:ca96c318ae2361da55903b8f6354020e686e1cbb068e7ae534285578b5ff9b2c +size 6162 diff --git a/excerpt-data/13d4b1406e769b80/gnomad.v4.0.sv.chr3.vcf.gz.tbi b/excerpt-data/13d4b1406e769b80/gnomad.v4.0.sv.chr3.vcf.gz.tbi deleted file mode 100644 index cad8f87..0000000 --- a/excerpt-data/13d4b1406e769b80/gnomad.v4.0.sv.chr3.vcf.gz.tbi +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:bb6fa1c42693ed4db554b389c579c70f1f29694d52db290d2718c5246068e5ef -size 170 diff --git a/excerpt-data/182806147755e799/gnomad.v4.0.sv.chr5.vcf.gz b/excerpt-data/182806147755e799/gnomad.v4.0.sv.chr5.vcf.gz index d4f3d44..e47118d 100644 --- a/excerpt-data/182806147755e799/gnomad.v4.0.sv.chr5.vcf.gz +++ b/excerpt-data/182806147755e799/gnomad.v4.0.sv.chr5.vcf.gz @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:43e5e11083b13f47017110cae922117c41ed43f9c8013e607bede09420f2e879 -size 416702 +oid sha256:c880a77429ad7b1dad3899ff22e4ed5b9faac3e28e267261ec619276d31d49fc +size 6164 diff --git a/excerpt-data/182806147755e799/gnomad.v4.0.sv.chr5.vcf.gz.tbi b/excerpt-data/182806147755e799/gnomad.v4.0.sv.chr5.vcf.gz.tbi deleted file mode 100644 index f1f4843..0000000 --- a/excerpt-data/182806147755e799/gnomad.v4.0.sv.chr5.vcf.gz.tbi +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:6080f8b3782bf82e3a8d94865db9e4c6f5a7a89fc5009fedcdbe307a20697dd4 -size 204 diff --git a/excerpt-data/19a12d845df10514/gnomad.v4.0.sv.chr18.vcf.gz b/excerpt-data/19a12d845df10514/gnomad.v4.0.sv.chr18.vcf.gz index 6ec4dbd..1ca9445 100644 --- a/excerpt-data/19a12d845df10514/gnomad.v4.0.sv.chr18.vcf.gz +++ b/excerpt-data/19a12d845df10514/gnomad.v4.0.sv.chr18.vcf.gz @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:34fa9e73c212aec43d742bef36eeae3d50b4997a17c5843e0ebf817c246bf6fc -size 381786 +oid sha256:bf9651ced472e66e0febe1171f483449015490d0b8a2a0e5b0305e9bc42a39a2 +size 6158 diff --git a/excerpt-data/19a12d845df10514/gnomad.v4.0.sv.chr18.vcf.gz.tbi b/excerpt-data/19a12d845df10514/gnomad.v4.0.sv.chr18.vcf.gz.tbi deleted file mode 100644 index 7ba3477..0000000 --- a/excerpt-data/19a12d845df10514/gnomad.v4.0.sv.chr18.vcf.gz.tbi +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:dc0fbd858f71ab97f0aad603982beeabb0edc6e830da0babfd8873d5e12a023a -size 212 diff --git a/excerpt-data/2245e9d640b3f206/gnomad.v4.0.sv.chrX.vcf.gz b/excerpt-data/2245e9d640b3f206/gnomad.v4.0.sv.chrX.vcf.gz index 4464d4b..c8ef0bd 100644 --- a/excerpt-data/2245e9d640b3f206/gnomad.v4.0.sv.chrX.vcf.gz +++ b/excerpt-data/2245e9d640b3f206/gnomad.v4.0.sv.chrX.vcf.gz @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:3127df711b9c38a08dc317be240704e3ef20b7b86e8e739677429823aa37fd4d -size 398015 +oid sha256:aeb14e798be93af8fba69a795200f2ff8b131d8fec894387fd22a525f9427a64 +size 6162 diff --git a/excerpt-data/2245e9d640b3f206/gnomad.v4.0.sv.chrX.vcf.gz.tbi b/excerpt-data/2245e9d640b3f206/gnomad.v4.0.sv.chrX.vcf.gz.tbi deleted file mode 100644 index 3ba8472..0000000 --- a/excerpt-data/2245e9d640b3f206/gnomad.v4.0.sv.chrX.vcf.gz.tbi +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:7073665b518fc929a8049a5b9101103ec7628f844d04ff8943ff98653ff91d59 -size 317 diff --git a/excerpt-data/3390d9ca9ce97bac/gnomad.v4.0.sv.chr2.vcf.gz b/excerpt-data/3390d9ca9ce97bac/gnomad.v4.0.sv.chr2.vcf.gz index c72d13f..7bc08cc 100644 --- a/excerpt-data/3390d9ca9ce97bac/gnomad.v4.0.sv.chr2.vcf.gz +++ b/excerpt-data/3390d9ca9ce97bac/gnomad.v4.0.sv.chr2.vcf.gz @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:d20cec346e12949e6eb0cb6b197063ce5d113b4a2d5e9073bc843274bec95aa5 -size 424653 +oid sha256:dedbbda1d6508b8a6e43c6d69ea8adbd0e3d359a65a2f34217192da74d717ff5 +size 6164 diff --git a/excerpt-data/3390d9ca9ce97bac/gnomad.v4.0.sv.chr2.vcf.gz.tbi b/excerpt-data/3390d9ca9ce97bac/gnomad.v4.0.sv.chr2.vcf.gz.tbi deleted file mode 100644 index 9c9387e..0000000 --- a/excerpt-data/3390d9ca9ce97bac/gnomad.v4.0.sv.chr2.vcf.gz.tbi +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:05175769d37521c3f7d755fe3adc6f575cec7b2e4799fb6806dca7f1a761fa66 -size 135 diff --git a/excerpt-data/3b19a5df654e6b4f/gnomad.v4.0.sv.chrY.vcf.gz b/excerpt-data/3b19a5df654e6b4f/gnomad.v4.0.sv.chrY.vcf.gz index 219880c..ded2f0b 100644 --- a/excerpt-data/3b19a5df654e6b4f/gnomad.v4.0.sv.chrY.vcf.gz +++ b/excerpt-data/3b19a5df654e6b4f/gnomad.v4.0.sv.chrY.vcf.gz @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:0b7ff9dc388be77a7f4ff603eec87d678d8a4ecd5913396988703e458cda553c -size 430809 +oid sha256:4fce3002b05ce4c7d4b4f62293ab09d9937a685de39d4123fdefdb08ab11f67c +size 6165 diff --git a/excerpt-data/3b19a5df654e6b4f/gnomad.v4.0.sv.chrY.vcf.gz.tbi b/excerpt-data/3b19a5df654e6b4f/gnomad.v4.0.sv.chrY.vcf.gz.tbi deleted file mode 100644 index 1d703af..0000000 --- a/excerpt-data/3b19a5df654e6b4f/gnomad.v4.0.sv.chrY.vcf.gz.tbi +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:a0e310746236a220fb0ba3d1dcbdeebc16499138d45f1ae25988d69371766ec5 -size 293 diff --git a/excerpt-data/41e368687b2c69e4/gnomad.v4.0.sv.chr7.vcf.gz b/excerpt-data/41e368687b2c69e4/gnomad.v4.0.sv.chr7.vcf.gz index 5c443d7..8a9d83a 100644 --- a/excerpt-data/41e368687b2c69e4/gnomad.v4.0.sv.chr7.vcf.gz +++ b/excerpt-data/41e368687b2c69e4/gnomad.v4.0.sv.chr7.vcf.gz @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:833c7e97efa6ecc0c2038c6bd9e07c63b1c26025e9efb39b599780c3ff00c2a0 -size 412586 +oid sha256:2b7775f9692dc0fc0c6dd341160d26ef0577b4e2ea3bd5e7fb4c3041b4bfb532 +size 6165 diff --git a/excerpt-data/41e368687b2c69e4/gnomad.v4.0.sv.chr7.vcf.gz.tbi b/excerpt-data/41e368687b2c69e4/gnomad.v4.0.sv.chr7.vcf.gz.tbi deleted file mode 100644 index 0cc454e..0000000 --- a/excerpt-data/41e368687b2c69e4/gnomad.v4.0.sv.chr7.vcf.gz.tbi +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:0517a6e670aeba67b7f449c7e55c52cb82560946996cf465c66034abb7e71fb7 -size 283 diff --git a/excerpt-data/555fd2fa82d1b73d/gnomad.v4.0.sv.chr11.vcf.gz b/excerpt-data/555fd2fa82d1b73d/gnomad.v4.0.sv.chr11.vcf.gz index 54cbe3b..fe31b69 100644 --- a/excerpt-data/555fd2fa82d1b73d/gnomad.v4.0.sv.chr11.vcf.gz +++ b/excerpt-data/555fd2fa82d1b73d/gnomad.v4.0.sv.chr11.vcf.gz @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:21a5dd8df2ba95bd07b6c237b3c11604e649b121973fd84e38565e0cb70dbcf0 -size 424520 +oid sha256:9d5cba684e7f602b5cd3973e225573b2b2dcb989797b14aef4a1e66316cbe41e +size 6159 diff --git a/excerpt-data/555fd2fa82d1b73d/gnomad.v4.0.sv.chr11.vcf.gz.tbi b/excerpt-data/555fd2fa82d1b73d/gnomad.v4.0.sv.chr11.vcf.gz.tbi deleted file mode 100644 index f389d98..0000000 --- a/excerpt-data/555fd2fa82d1b73d/gnomad.v4.0.sv.chr11.vcf.gz.tbi +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:43ef41f6c4cc59b8b188fa3822f3863661f840af87326aac623ab6155541161e -size 300 diff --git a/excerpt-data/6433022d4690f461/gnomad.v4.0.sv.chr21.vcf.gz b/excerpt-data/6433022d4690f461/gnomad.v4.0.sv.chr21.vcf.gz index 637fcc5..1ebd65d 100644 --- a/excerpt-data/6433022d4690f461/gnomad.v4.0.sv.chr21.vcf.gz +++ b/excerpt-data/6433022d4690f461/gnomad.v4.0.sv.chr21.vcf.gz @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:c1aea4d0f2dac8fe5c0101bb8d4b0284f03720a014776ecb448e6ab0e0d2c51d -size 403589 +oid sha256:9793eb66c2bbc2f32459316ea2259c8000845a424a0a8c0ad7d243323083542c +size 6160 diff --git a/excerpt-data/6433022d4690f461/gnomad.v4.0.sv.chr21.vcf.gz.tbi b/excerpt-data/6433022d4690f461/gnomad.v4.0.sv.chr21.vcf.gz.tbi deleted file mode 100644 index cd03395..0000000 --- a/excerpt-data/6433022d4690f461/gnomad.v4.0.sv.chr21.vcf.gz.tbi +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:2b9b5e428fdbe544a37f0db1f63ebf7322e5847210be9e2f0995d44753020878 -size 274 diff --git a/excerpt-data/6f337fba39cc4c2e/gnomad.v4.0.sv.chr13.vcf.gz b/excerpt-data/6f337fba39cc4c2e/gnomad.v4.0.sv.chr13.vcf.gz index e7fbdcc..5dd41e2 100644 --- a/excerpt-data/6f337fba39cc4c2e/gnomad.v4.0.sv.chr13.vcf.gz +++ b/excerpt-data/6f337fba39cc4c2e/gnomad.v4.0.sv.chr13.vcf.gz @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:ae4baafd36a434434e3dd8a9e66a1a25387e4ca1c486e7b2d17f0ed6301187a5 -size 513889 +oid sha256:55c9dd0187aecfb89ee92d463bf90a1d1a9041471bb8adf03c25bf3a4086ee97 +size 6158 diff --git a/excerpt-data/6f337fba39cc4c2e/gnomad.v4.0.sv.chr13.vcf.gz.tbi b/excerpt-data/6f337fba39cc4c2e/gnomad.v4.0.sv.chr13.vcf.gz.tbi deleted file mode 100644 index b44c5ac..0000000 --- a/excerpt-data/6f337fba39cc4c2e/gnomad.v4.0.sv.chr13.vcf.gz.tbi +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:f9c1e4e65c1abd5f78ef55075c3d81e94bb5d074310adfabff3f4ddd42d79b40 -size 297 diff --git a/excerpt-data/9cc0f2ee73775510/gnomad.v4.0.sv.chr10.vcf.gz b/excerpt-data/9cc0f2ee73775510/gnomad.v4.0.sv.chr10.vcf.gz index 6969b91..7feb43b 100644 --- a/excerpt-data/9cc0f2ee73775510/gnomad.v4.0.sv.chr10.vcf.gz +++ b/excerpt-data/9cc0f2ee73775510/gnomad.v4.0.sv.chr10.vcf.gz @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:32b8f2049412947a837c1dc5d04a55dca2cbf31cd22e899b6a297a4752c5fdfa -size 390567 +oid sha256:941a1c493771251e3fbed3a55a054a2d1a84ccafb678299eb5845f54ac001542 +size 6286 diff --git a/excerpt-data/9cc0f2ee73775510/gnomad.v4.0.sv.chr10.vcf.gz.tbi b/excerpt-data/9cc0f2ee73775510/gnomad.v4.0.sv.chr10.vcf.gz.tbi deleted file mode 100644 index 8011123..0000000 --- a/excerpt-data/9cc0f2ee73775510/gnomad.v4.0.sv.chr10.vcf.gz.tbi +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:86992343986e8d296dba0980b53dc3622f24d1c8521b08385e47c5aece22a75a -size 213 diff --git a/excerpt-data/a271968c7bf845c0/gnomad.v4.0.sv.chr16.vcf.gz b/excerpt-data/a271968c7bf845c0/gnomad.v4.0.sv.chr16.vcf.gz index 322d0da..c9a44ab 100644 --- a/excerpt-data/a271968c7bf845c0/gnomad.v4.0.sv.chr16.vcf.gz +++ b/excerpt-data/a271968c7bf845c0/gnomad.v4.0.sv.chr16.vcf.gz @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:68b2163951986ea1206924defa2e9a22ca31f7a5bb465e9b1a29fbbb584ec047 -size 388996 +oid sha256:a0e890d7aa10cb54ec517c472fa06e279a56b181092a2d775f03cf071fa38fab +size 6161 diff --git a/excerpt-data/a271968c7bf845c0/gnomad.v4.0.sv.chr16.vcf.gz.tbi b/excerpt-data/a271968c7bf845c0/gnomad.v4.0.sv.chr16.vcf.gz.tbi deleted file mode 100644 index 8d58c7f..0000000 --- a/excerpt-data/a271968c7bf845c0/gnomad.v4.0.sv.chr16.vcf.gz.tbi +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:1a5db430f4f13d480b945a2c2c737d6f64b1f44d3910b66e13a9077b7a608d38 -size 289 diff --git a/excerpt-data/a83da02295a7ae02/gnomad.v4.0.sv.chr20.vcf.gz b/excerpt-data/a83da02295a7ae02/gnomad.v4.0.sv.chr20.vcf.gz index e9079b7..d404c5f 100644 --- a/excerpt-data/a83da02295a7ae02/gnomad.v4.0.sv.chr20.vcf.gz +++ b/excerpt-data/a83da02295a7ae02/gnomad.v4.0.sv.chr20.vcf.gz @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:45a1c85a7356d6221410ad7aefeeef836403758cfe2bceebc0cee4c0987b3554 -size 376158 +oid sha256:5e936a8dd4e6e22351380963e90ef833645ac925f8cf3d2ce45203adfe3651ac +size 6159 diff --git a/excerpt-data/a83da02295a7ae02/gnomad.v4.0.sv.chr20.vcf.gz.tbi b/excerpt-data/a83da02295a7ae02/gnomad.v4.0.sv.chr20.vcf.gz.tbi deleted file mode 100644 index ad5e235..0000000 --- a/excerpt-data/a83da02295a7ae02/gnomad.v4.0.sv.chr20.vcf.gz.tbi +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:19aeacefa75b206ec58916e13cc2f84ea7c2c870db0fa9544d85e7d5c224fd29 -size 239 diff --git a/excerpt-data/aea53328ecd0b712/gnomad.v4.0.sv.chr6.vcf.gz b/excerpt-data/aea53328ecd0b712/gnomad.v4.0.sv.chr6.vcf.gz index f3ddf41..6a21b8b 100644 --- a/excerpt-data/aea53328ecd0b712/gnomad.v4.0.sv.chr6.vcf.gz +++ b/excerpt-data/aea53328ecd0b712/gnomad.v4.0.sv.chr6.vcf.gz @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:11e3601dec127e533cc7902490cb4c0c29a002c85e6bf157200409fbdc392597 -size 396393 +oid sha256:5d6ea9137a0181b7d7dbf6a6bf2ee6dc7b6f95582f32a1df4b9a3e8dfdb22f35 +size 6162 diff --git a/excerpt-data/aea53328ecd0b712/gnomad.v4.0.sv.chr6.vcf.gz.tbi b/excerpt-data/aea53328ecd0b712/gnomad.v4.0.sv.chr6.vcf.gz.tbi deleted file mode 100644 index a62c841..0000000 --- a/excerpt-data/aea53328ecd0b712/gnomad.v4.0.sv.chr6.vcf.gz.tbi +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:51b989744861351f01c9a9ee9074e93d257925bf5552b94fdb00110e4c82c8e7 -size 270 diff --git a/excerpt-data/bc6f1860df305e83/gnomad.v4.0.sv.chr17.vcf.gz b/excerpt-data/bc6f1860df305e83/gnomad.v4.0.sv.chr17.vcf.gz index 0526ba5..70df8d1 100644 --- a/excerpt-data/bc6f1860df305e83/gnomad.v4.0.sv.chr17.vcf.gz +++ b/excerpt-data/bc6f1860df305e83/gnomad.v4.0.sv.chr17.vcf.gz @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:fd6c9d5dd5475c57b5da8bc2d60ad97ba08769796cfba18ceaf31e12bf97700f -size 389157 +oid sha256:7e327e8202e93f69005e717037e2664018c40220508935819a5bed1bcbddbbb8 +size 6159 diff --git a/excerpt-data/bc6f1860df305e83/gnomad.v4.0.sv.chr17.vcf.gz.tbi b/excerpt-data/bc6f1860df305e83/gnomad.v4.0.sv.chr17.vcf.gz.tbi deleted file mode 100644 index c05e188..0000000 --- a/excerpt-data/bc6f1860df305e83/gnomad.v4.0.sv.chr17.vcf.gz.tbi +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:8543baabca5fd4e7b31bf104ca3933e033feeaa80935df3308664018c4255596 -size 307 diff --git a/excerpt-data/c3ce41a42c6319e0/gnomad.v4.0.sv.chr22.vcf.gz b/excerpt-data/c3ce41a42c6319e0/gnomad.v4.0.sv.chr22.vcf.gz index bafb9c2..bc455c2 100644 --- a/excerpt-data/c3ce41a42c6319e0/gnomad.v4.0.sv.chr22.vcf.gz +++ b/excerpt-data/c3ce41a42c6319e0/gnomad.v4.0.sv.chr22.vcf.gz @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:f06299de29069f7bb1354720250079f0bd691376ad9274f1ecd42e02b33a9d26 -size 407737 +oid sha256:12dbabb41eed02b053fc6d7b4a3a977a34ae441efe91449fe5bc1a145824b780 +size 6159 diff --git a/excerpt-data/c3ce41a42c6319e0/gnomad.v4.0.sv.chr22.vcf.gz.tbi b/excerpt-data/c3ce41a42c6319e0/gnomad.v4.0.sv.chr22.vcf.gz.tbi deleted file mode 100644 index b86c2e1..0000000 --- a/excerpt-data/c3ce41a42c6319e0/gnomad.v4.0.sv.chr22.vcf.gz.tbi +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:c89d49b0895ac0624702ecb5cf3ad9aaf3583c553dda1461ec7db5f35b9d9d24 -size 247 diff --git a/excerpt-data/ca218c36ad8374d2/gnomad.v4.0.sv.chr14.vcf.gz b/excerpt-data/ca218c36ad8374d2/gnomad.v4.0.sv.chr14.vcf.gz index 2b4f09a..e189c3f 100644 --- a/excerpt-data/ca218c36ad8374d2/gnomad.v4.0.sv.chr14.vcf.gz +++ b/excerpt-data/ca218c36ad8374d2/gnomad.v4.0.sv.chr14.vcf.gz @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:9680b2acae61e70b665a4b4ce01aa80450fa40b349c73ea86220857242bc69c9 -size 390266 +oid sha256:39427cd58bb32c705006424be9138271e570f8c00a77abdaae7b223020152d5d +size 6158 diff --git a/excerpt-data/ca218c36ad8374d2/gnomad.v4.0.sv.chr14.vcf.gz.tbi b/excerpt-data/ca218c36ad8374d2/gnomad.v4.0.sv.chr14.vcf.gz.tbi deleted file mode 100644 index a6a7b00..0000000 --- a/excerpt-data/ca218c36ad8374d2/gnomad.v4.0.sv.chr14.vcf.gz.tbi +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:0ed6a3bf8df7d6bbe8a9ec03e872a27fe3dfb83c45334afa9de77979a35d7503 -size 274 diff --git a/excerpt-data/ec5c8ad8a9c29dd0/gnomad.v4.0.sv.chr12.vcf.gz b/excerpt-data/ec5c8ad8a9c29dd0/gnomad.v4.0.sv.chr12.vcf.gz index 4d7d200..51c0f09 100644 --- a/excerpt-data/ec5c8ad8a9c29dd0/gnomad.v4.0.sv.chr12.vcf.gz +++ b/excerpt-data/ec5c8ad8a9c29dd0/gnomad.v4.0.sv.chr12.vcf.gz @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:c83d34bfb14dabcb882ead05e1f4d93d21570d7323a924576577723d33f83368 -size 394149 +oid sha256:247152b8b0b1c32f65020f4c01273bfe11ad9430a4320ceccc64b8c98187b513 +size 6161 diff --git a/excerpt-data/ec5c8ad8a9c29dd0/gnomad.v4.0.sv.chr12.vcf.gz.tbi b/excerpt-data/ec5c8ad8a9c29dd0/gnomad.v4.0.sv.chr12.vcf.gz.tbi deleted file mode 100644 index 8978e7c..0000000 --- a/excerpt-data/ec5c8ad8a9c29dd0/gnomad.v4.0.sv.chr12.vcf.gz.tbi +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:a1f45cc748138b3fe02e88d974183023786371e1bbc3a0c21516790fd1436405 -size 197 diff --git a/excerpt-data/eed573336b6205b5/gnomad.v4.0.sv.chr19.vcf.gz b/excerpt-data/eed573336b6205b5/gnomad.v4.0.sv.chr19.vcf.gz index dcd5648..8b9ce0c 100644 --- a/excerpt-data/eed573336b6205b5/gnomad.v4.0.sv.chr19.vcf.gz +++ b/excerpt-data/eed573336b6205b5/gnomad.v4.0.sv.chr19.vcf.gz @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:278c849bcd5536c270761a8c9546c1669074bbc3a07630445096de163cf90155 -size 407544 +oid sha256:dda56b56ff439a9b842f151623615e37367850383a185dc41106ca033b5e68e9 +size 6158 diff --git a/excerpt-data/eed573336b6205b5/gnomad.v4.0.sv.chr19.vcf.gz.tbi b/excerpt-data/eed573336b6205b5/gnomad.v4.0.sv.chr19.vcf.gz.tbi deleted file mode 100644 index e510d68..0000000 --- a/excerpt-data/eed573336b6205b5/gnomad.v4.0.sv.chr19.vcf.gz.tbi +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:d320c44686ec998acba35fd29dc265738b551b94dff4f12039a46229841e6d15 -size 212 diff --git a/excerpt-data/ef1e660e8e0359bc/gnomad.v4.0.sv.chr1.vcf.gz b/excerpt-data/ef1e660e8e0359bc/gnomad.v4.0.sv.chr1.vcf.gz index e4a5ee8..020716f 100644 --- a/excerpt-data/ef1e660e8e0359bc/gnomad.v4.0.sv.chr1.vcf.gz +++ b/excerpt-data/ef1e660e8e0359bc/gnomad.v4.0.sv.chr1.vcf.gz @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:94d8ed19180abf8922f5b863ae3ce8bffd9e186aebadd8e81bb3c88eaeef5709 -size 430679 +oid sha256:a80d59579012e9cc4f17971960196e7db8ade9e5df1cf74c860f49fdae897245 +size 6272 diff --git a/excerpt-data/ef1e660e8e0359bc/gnomad.v4.0.sv.chr1.vcf.gz.tbi b/excerpt-data/ef1e660e8e0359bc/gnomad.v4.0.sv.chr1.vcf.gz.tbi deleted file mode 100644 index 48ced38..0000000 --- a/excerpt-data/ef1e660e8e0359bc/gnomad.v4.0.sv.chr1.vcf.gz.tbi +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:fb12c4f84202a4839fed7c6f581f8ee44a14aedfea5d4368cf7eb0d34d9d2b5c -size 206 diff --git a/excerpt-data/f5375ffc6d0e6adb/gnomad.v4.0.sv.chr9.vcf.gz b/excerpt-data/f5375ffc6d0e6adb/gnomad.v4.0.sv.chr9.vcf.gz index c901c95..459f683 100644 --- a/excerpt-data/f5375ffc6d0e6adb/gnomad.v4.0.sv.chr9.vcf.gz +++ b/excerpt-data/f5375ffc6d0e6adb/gnomad.v4.0.sv.chr9.vcf.gz @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:a6c16bcd58417d4711d91dc190ea44c667ebc07f59c71af3418b94ba22a68263 -size 426303 +oid sha256:c53d69003c0961136dfed2d33aa5da20733b7529fd57c2bef661a3f77b0a517f +size 6161 diff --git a/excerpt-data/f5375ffc6d0e6adb/gnomad.v4.0.sv.chr9.vcf.gz.tbi b/excerpt-data/f5375ffc6d0e6adb/gnomad.v4.0.sv.chr9.vcf.gz.tbi deleted file mode 100644 index b2b0386..0000000 --- a/excerpt-data/f5375ffc6d0e6adb/gnomad.v4.0.sv.chr9.vcf.gz.tbi +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:a53e0f00761d5178b0efa713bb8b761a0032089d99c055b0da7a2331c4d5bbc5 -size 234 diff --git a/excerpt-data/fe2a55106424e727/gnomad.v4.0.sv.chr8.vcf.gz b/excerpt-data/fe2a55106424e727/gnomad.v4.0.sv.chr8.vcf.gz index 91548ba..5af663c 100644 --- a/excerpt-data/fe2a55106424e727/gnomad.v4.0.sv.chr8.vcf.gz +++ b/excerpt-data/fe2a55106424e727/gnomad.v4.0.sv.chr8.vcf.gz @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:d59084887e82047051f15439598bdad4aa2d00052e01a01a0b8ed3caff80a38b -size 408984 +oid sha256:5a2089c979f86c293d0d7ac97f85505c6a8dfa0f7b31729219e4fd9dbbe7ef8a +size 6164 diff --git a/excerpt-data/fe2a55106424e727/gnomad.v4.0.sv.chr8.vcf.gz.tbi b/excerpt-data/fe2a55106424e727/gnomad.v4.0.sv.chr8.vcf.gz.tbi deleted file mode 100644 index c3cb02d..0000000 --- a/excerpt-data/fe2a55106424e727/gnomad.v4.0.sv.chr8.vcf.gz.tbi +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:9073baa6b820ce0b318e81d5d7e9682a615472336222ab4659e4376a8688d8fe -size 235 diff --git a/rules/output/worker/bgdb.smk b/rules/output/worker/bgdb.smk index 1fcf0d0..3caf612 100644 --- a/rules/output/worker/bgdb.smk +++ b/rules/output/worker/bgdb.smk @@ -9,10 +9,10 @@ rule output_worker_bgdb_g1k: spec=f"output/full/worker/bgdb-g1k-grch37-{{version}}+{PV.worker}/bgdb-g1k.spec.yaml", shell: r""" - varfish-server-worker db to-bin \ + varfish-server-worker strucvars txt-to-bin \ --input-type strucvar-g1k \ --path-input {input.bed} \ - --path-output-bin {output.bin} + --path-output {output.bin} varfish-db-downloader tpl \ --template rules/output/worker/bgdb.spec.yaml \ @@ -40,10 +40,10 @@ rule output_worker_bgdb_exac: spec=f"output/full/worker/bgdb-exac-grch37-{{version}}+{PV.worker}/bgdb-exac.spec.yaml", shell: r""" - varfish-server-worker db to-bin \ + varfish-server-worker strucvars txt-to-bin \ --input-type strucvar-exac-cnv \ --path-input {input.bed} \ - --path-output-bin {output.bin} + --path-output {output.bin} varfish-db-downloader tpl \ --template rules/output/worker/bgdb.spec.yaml \ @@ -63,7 +63,38 @@ rule output_worker_bgdb_exac: """ -rule output_worker_bgdb_gnomad: +rule output_worker_bgdb_gnomad_exomes_cnv_grch38: + input: + bed=f"output/full/tracks/track-strucvars-gnomad-cnv-grch38-{{version}}+{DV.tracks}/gnomad-cnv.bed.gz", + output: + bin=f"output/full/worker/bgdb-gnomad-exomes-cnv-grch38-{{version}}+{PV.worker}/bgdb-gnomad-exomes-cnv-grch38.bin", + spec=f"output/full/worker/bgdb-gnomad-exomes-cnv-grch38-{{version}}+{PV.worker}/bgdb-gnomad-exomes-cnv-grch38.spec.yaml", + shell: + r""" + varfish-server-worker strucvars txt-to-bin \ + --input-type strucvar-gnomad-cnv4 \ + --path-input {input.bed} \ + --path-output {output.bin} + + varfish-db-downloader tpl \ + --template rules/output/worker/bgdb.spec.yaml \ + \ + --value db_name=gnomad-exomes-cnv \ + --value title="gnomAD Exomes CNV" \ + --value creator="gnomAD Consortium" \ + --value source="https://gnomad.broadinstitute.org/downloads#v4-copy-number-variants" \ + \ + --value version={wildcards.version}+{PV.worker} \ + --value today={TODAY} \ + --value genome_release=grch38 \ + \ + --value v_worker={PV.worker} \ + --value v_downloader={PV.downloader} \ + > {output.spec} + """ + + +rule output_worker_bgdb_gnomad_sv_grch37: input: bed=f"output/full/tracks/track-strucvars-gnomad-grch37-{{version}}+{DV.tracks}/gnomad.bed.gz", output: @@ -71,11 +102,10 @@ rule output_worker_bgdb_gnomad: spec=f"output/full/worker/bgdb-gnomad-grch37-{{version}}+{PV.worker}/bgdb-gnomad.spec.yaml", shell: r""" - varfish-server-worker db to-bin \ - --input-type strucvar-gnomad-sv \ + varfish-server-worker strucvars txt-to-bin \ + --input-type strucvar-gnomad-sv2 \ --path-input {input.bed} \ - --path-output-bin {output.bin} - + --path-output {output.bin} varfish-db-downloader tpl \ --template rules/output/worker/bgdb.spec.yaml \ @@ -95,6 +125,37 @@ rule output_worker_bgdb_gnomad: """ +rule output_worker_bgdb_gnomad_genomes_sv_grch38: + input: + bed=f"output/full/tracks/track-strucvars-gnomad-sv-grch38-{{version}}+{DV.tracks}/gnomad-sv.bed.gz", + output: + bin=f"output/full/worker/bgdb-gnomad-genomes-sv-grch38-{{version}}+{PV.worker}/bgdb-gnomad-genomes-sv-grch38.bin", + spec=f"output/full/worker/bgdb-gnomad-genomes-sv-grch38-{{version}}+{PV.worker}/bgdb-gnomad-genomes-sv-grch38.spec.yaml", + shell: + r""" + varfish-server-worker strucvars txt-to-bin \ + --input-type strucvar-gnomad-sv4 \ + --path-input {input.bed} \ + --path-output {output.bin} + + varfish-db-downloader tpl \ + --template rules/output/worker/bgdb.spec.yaml \ + \ + --value db_name=gnomad \ + --value title="gnomAD-SVs" \ + --value creator="gnomAD Consortium" \ + --value source="https://gnomad.broadinstitute.org/downloads#v4-structural-variants" \ + \ + --value version={wildcards.version}+{PV.worker} \ + --value today={TODAY} \ + --value genome_release=grch38 \ + \ + --value v_worker={PV.worker} \ + --value v_downloader={PV.downloader} \ + > {output.spec} + """ + + rule output_worker_bgdb_dbvar: input: bed=f"output/full/tracks/track-strucvars-dbvar-{{genome_release}}-{{version}}+{DV.tracks}/dbvar.bed.gz", @@ -103,10 +164,10 @@ rule output_worker_bgdb_dbvar: spec=f"output/full/worker/bgdb-dbvar-{{genome_release}}-{{version}}+{PV.worker}/bgdb-dbvar.spec.yaml", shell: r""" - varfish-server-worker db to-bin \ + varfish-server-worker strucvars txt-to-bin \ --input-type strucvar-db-var \ --path-input {input.bed} \ - --path-output-bin {output.bin} + --path-output {output.bin} varfish-db-downloader tpl \ --template rules/output/worker/bgdb.spec.yaml \ @@ -134,10 +195,10 @@ rule output_worker_bgdb_dgv: spec=f"output/full/worker/bgdb-dgv-{{genome_release}}-{{version}}+{PV.worker}/bgdb-dgv.spec.yaml", shell: r""" - varfish-server-worker db to-bin \ + varfish-server-worker strucvars txt-to-bin \ --input-type strucvar-dgv \ --path-input {input.bed} \ - --path-output-bin {output.bin} + --path-output {output.bin} varfish-db-downloader tpl \ --template rules/output/worker/bgdb.spec.yaml \ @@ -165,10 +226,10 @@ rule output_worker_bgdb_dgv_gs: spec=f"output/full/worker/bgdb-dgv-gs-{{genome_release}}-{{version}}+{PV.worker}/bgdb-dgv-gs.spec.yaml", shell: r""" - varfish-server-worker db to-bin \ + varfish-server-worker strucvars txt-to-bin \ --input-type strucvar-dgv-gs \ --path-input {input.bed} \ - --path-output-bin {output.bin} + --path-output {output.bin} varfish-db-downloader tpl \ --template rules/output/worker/bgdb.spec.yaml \ diff --git a/rules/output/worker/clinvar.smk b/rules/output/worker/clinvar.smk index 3230ab5..931d53c 100644 --- a/rules/output/worker/clinvar.smk +++ b/rules/output/worker/clinvar.smk @@ -26,7 +26,7 @@ rule annos_strucvar_clinvar_convert: varfish-server-worker db to-bin \ --input-type clinvar-sv \ --path-input {input.tsv} \ - --path-output-bin {output.bin} + --path-output {output.bin} varfish-db-downloader tpl \ --template rules/output/worker/clinvar_strucvars.spec.yaml \ diff --git a/rules/output/worker/genes_regions.smk b/rules/output/worker/genes_regions.smk index c7cd499..03685c3 100644 --- a/rules/output/worker/genes_regions.smk +++ b/rules/output/worker/genes_regions.smk @@ -20,7 +20,7 @@ rule genes_regions_worker_convert: db to-bin \ --input-type gene-region \ --path-input {input.bed} \ - --path-output-bin {output.bin} + --path-output {output.bin} varfish-db-downloader tpl \ --template rules/output/worker/{wildcards.source}_genes.spec.yaml \ diff --git a/rules/output/worker/hgnc.smk b/rules/output/worker/hgnc.smk index 1babd54..6f86df5 100644 --- a/rules/output/worker/hgnc.smk +++ b/rules/output/worker/hgnc.smk @@ -17,7 +17,7 @@ rule output_hgnc_xlink_binary: varfish-server-worker db to-bin \ --input-type xlink \ --path-input {input.tsv} \ - --path-output-bin {output.bin} + --path-output {output.bin} varfish-db-downloader tpl \ --template rules/output/worker/hgnc_xlink.spec.yaml \ diff --git a/rules/output/worker/masked.smk b/rules/output/worker/masked.smk index 4e46f74..71a6124 100644 --- a/rules/output/worker/masked.smk +++ b/rules/output/worker/masked.smk @@ -15,7 +15,7 @@ rule output_masked_repeat: varfish-server-worker db to-bin \ --input-type masked-region \ --path-input {input} \ - --path-output-bin {output.bin} + --path-output {output.bin} varfish-db-downloader tpl \ --template rules/output/worker/masked_repeat.spec.yaml \ @@ -44,7 +44,7 @@ rule output_masked_segdup: varfish-server-worker db to-bin \ --input-type masked-region \ --path-input {input} \ - --path-output-bin {output.bin} + --path-output {output.bin} varfish-db-downloader tpl \ --template rules/output/worker/masked_segdup.spec.yaml \ diff --git a/rules/work/annos/strucvars/gnomad.smk b/rules/work/annos/strucvars/gnomad.smk index 33467f7..6e824fd 100644 --- a/rules/work/annos/strucvars/gnomad.smk +++ b/rules/work/annos/strucvars/gnomad.smk @@ -1,7 +1,7 @@ ## Rules related to gnomAD-SV. -rule annos_strucvars_gnomad_grch37_download: # -- download gnomAD-SV files +rule annos_strucvars_gnomad_sv2_grch37_download: # -- download gnomAD-SV v2 files output: vcf="work/download/annos/grch37/strucvars/gnomad/2.1.1/gnomad_v2.1_sv.{token}.vcf.gz", shell: @@ -12,7 +12,7 @@ rule annos_strucvars_gnomad_grch37_download: # -- download gnomAD-SV files """ -rule annos_strucvars_gnomad_grch37_process: # -- process gnomAD-SV files +rule annos_strucvars_gnomad_sv2_grch37_process: # -- process gnomAD-SV v2 files input: vcf="work/download/annos/grch37/strucvars/gnomad/2.1.1/gnomad_v2.1_sv.sites.vcf.gz", output: diff --git a/rules/work/annos/strucvars/gnomad_sv4.smk b/rules/work/annos/strucvars/gnomad_sv4.smk index bf7ef03..b47d9b5 100644 --- a/rules/work/annos/strucvars/gnomad_sv4.smk +++ b/rules/work/annos/strucvars/gnomad_sv4.smk @@ -12,6 +12,64 @@ rule annos_strucvars_gnomad_sv_4_grch38_download: # -- download gnomAD-SV 4.0 f """ +rule annos_strucvars_gnomad_sv_4_grch38_process: # -- process gnomAD-SV v4 files + input: + vcf=[ + f"work/download/annos/grch38/strucvars/gnomad_sv/{{version}}/gnomad.v{{version}}.sv.chr{chrom}.vcf.gz" + for chrom in list(range(1, 23)) + ["X", "Y"] + ], + output: + bed=f"output/full/tracks/track-strucvars-gnomad-sv-grch38-{{version}}+{DV.tracks}/gnomad-sv.bed.gz", + bed_md5=f"output/full/tracks/track-strucvars-gnomad-sv-grch38-{{version}}+{DV.tracks}/gnomad-sv.bed.gz.md5", + bed_tbi=f"output/full/tracks/track-strucvars-gnomad-sv-grch38-{{version}}+{DV.tracks}/gnomad-sv.bed.gz.tbi", + bed_tbi_md5=f"output/full/tracks/track-strucvars-gnomad-sv-grch38-{{version}}+{DV.tracks}/gnomad-sv.bed.gz.tbi.md5", + shell: + r""" + export TMPDIR=$(mktemp -d) + trap "rm -rf $TMPDIR" ERR EXIT + + echo -e "#chromosome\tbegin\tend\tsv_type\tmale_n_homref\tmale_n_het\tmale_n_homalt\tmale_n_hemiref\tmale_n_hemialt\tfemale_n_homref\tfemale_n_het\tfemale_n_homalt\tcnv_n_total\tcnv_n_var" \ + > $TMPDIR/tmp.bed + + for vcf in {input.vcf}; do + bcftools query \ + -f "%CHROM\t%POS0\t%INFO/END\t%INFO/SVTYPE\t%INFO/MALE_N_HOMREF\t%INFO/MALE_N_HET\t%INFO/MALE_N_HOMALT\t%INFO/MALE_N_HEMIREF\t%INFO/MALE_N_HEMIALT\t%INFO/FEMALE_N_HOMREF\t%INFO/FEMALE_N_HET\t%INFO/FEMALE_N_HOMALT\t%CN_NUMBER\t%CN_COUNT\n" \ + $vcf \ + | awk -v OFS='\t' '{{ + if ($5 == ".") {{ $5 = 0; }} + if ($6 == ".") {{ $6 = 0; }} + if ($7 == ".") {{ $7 = 0; }} + if ($8 == ".") {{ $8 = 0; }} + if ($9 == ".") {{ $9 = 0; }} + if ($10 == ".") {{ $10 = 0; }} + if ($11 == ".") {{ $11 = 0; }} + if ($12 == ".") {{ $12 = 0; }} + if ($13 == ".") {{ $13 = 0; }} + if ($14 == ".") {{ + $14 = 0 + }} else {{ + sum = 0 + split($14, a, ",") + for (x in a) {{ + sum += x + }} + $14 = sum + }} + print $0 + }}' \ + | sed -e 's/CPX/BND/g' -e 's/CTX/BND/g' \ + >> $TMPDIR/tmp.bed + done + + bgzip -c $TMPDIR/tmp.bed >{output.bed} + + tabix -p bed -S 1 -f {output.bed} + + md5sum {output.bed} >{output.bed_md5} + md5sum {output.bed_tbi} >{output.bed_tbi_md5} + """ + + rule annos_strucvars_gnomad_cnv_4_grch38_download: # -- download gnomAD-CNV 4.0 files output: vcf="work/download/annos/grch38/strucvars/gnomad_cnv/4.0/gnomad.v4.0.cnv.{token}.vcf.gz", @@ -21,3 +79,35 @@ rule annos_strucvars_gnomad_cnv_4_grch38_download: # -- download gnomAD-CNV 4.0 -O {output.vcf} \ https://storage.googleapis.com/gcp-public-data--gnomad/release/4.0/exome_cnv/gnomad.v4.0.cnv.{wildcards.token}.vcf.gz """ + + +rule annos_strucvars_gnomad_cnv_4_grch38_process: # -- process gnomAD-CNV 4.0 files + input: + vcf="work/download/annos/grch38/strucvars/gnomad_cnv/{version}/gnomad.v{version}.cnv.all.vcf.gz", + output: + bed=f"output/full/tracks/track-strucvars-gnomad-cnv-grch38-{{version}}+{DV.tracks}/gnomad-cnv.bed.gz", + bed_md5=f"output/full/tracks/track-strucvars-gnomad-cnv-grch38-{{version}}+{DV.tracks}/gnomad-cnv.bed.gz.md5", + bed_tbi=f"output/full/tracks/track-strucvars-gnomad-cnv-grch38-{{version}}+{DV.tracks}/gnomad-cnv.bed.gz.tbi", + bed_tbi_md5=f"output/full/tracks/track-strucvars-gnomad-cnv-grch38-{{version}}+{DV.tracks}/gnomad-cnv.bed.gz.tbi.md5", + shell: + r""" + set -euo pipefail + + export TMPDIR=$(mktemp -d) + trap "rm -rf $TMPDIR" ERR EXIT + + echo -e "#chromosome\tbegin\tend\tsv_type\tn_total\tn_var" \ + > $TMPDIR/tmp.bed + + bcftools query \ + -f "%CHROM\t%POS0\t%INFO/END\t%INFO/SVTYPE\t%INFO/SN\t%INFO/SC\n" \ + {input.vcf} \ + >> $TMPDIR/tmp.bed + + bgzip -c $TMPDIR/tmp.bed >{output.bed} + + tabix -p bed -S 1 -f {output.bed} + + md5sum {output.bed} >{output.bed_md5} + md5sum {output.bed_tbi} >{output.bed_tbi_md5} + """