feat: write out fileDate on ingest (#210) (#215)

varfish-org · Oct 10, 2023 · 7731958 · 7731958
1 parent d6db0b7
commit 7731958
Show file tree

Hide file tree

Showing 18 changed files with 497 additions and 468 deletions.
diff --git a/src/seqvars/ingest/header.rs b/src/seqvars/ingest/header.rs
@@ -182,6 +182,7 @@ pub fn build_output_header(
     input_header: &vcf::Header,
     pedigree: &Option<mehari::ped::PedigreeByName>,
     genomebuild: GenomeRelease,
+    file_date: &str,
     case_uuid: &uuid::Uuid,
     worker_version: &str,
 ) -> Result<vcf::Header, anyhow::Error> {
@@ -193,6 +194,10 @@ pub fn build_output_header(
     use vcf::record::genotypes::keys::key;
 
     let builder = vcf::Header::builder()
+        .insert(
+            "fileDate".parse()?,
+            vcf::header::record::Value::from(file_date),
+        )?
         .add_filter("PASS", Map::<Filter>::new("All filters passed"))
         .add_info(
             "gnomad_exomes_an".parse()?,
@@ -466,6 +471,7 @@ mod test {
             &input_vcf_header,
             &Some(pedigree),
             crate::common::GenomeRelease::Grch37,
+            "20230421",
             &uuid::Uuid::parse_str("00000000-0000-0000-0000-000000000000").unwrap(),
             "x.y.z",
         )?;
@@ -500,6 +506,7 @@ mod test {
             &input_vcf_header,
             &Some(pedigree),
             crate::common::GenomeRelease::Grch38,
+            "20230421",
             &uuid::Uuid::parse_str("00000000-0000-0000-0000-000000000000").unwrap(),
             "x.y.z",
         )?;

diff --git a/src/seqvars/ingest/mod.rs b/src/seqvars/ingest/mod.rs
@@ -14,15 +14,19 @@ pub mod header;
 #[derive(Debug, clap::Parser)]
 #[command(author, version, about = "ingest sequence variant VCF", long_about = None)]
 pub struct Args {
-    /// The path to the mehari database.
-    #[clap(long)]
-    pub path_mehari_db: String,
+    /// Value to write to `##fileDate`.
+    #[arg(long)]
+    pub file_date: String,
     /// The case UUID to write out.
     #[clap(long)]
     pub case_uuid: uuid::Uuid,
     /// The assumed genome build.
     #[clap(long)]
     pub genomebuild: GenomeRelease,
+
+    /// The path to the mehari database.
+    #[clap(long)]
+    pub path_mehari_db: String,
     /// Path to the pedigree file.
     #[clap(long)]
     pub path_ped: String,
@@ -445,6 +449,7 @@ pub fn run(args_common: &crate::common::Args, args: &Args) -> Result<(), anyhow:
         &input_header,
         &Some(pedigree),
         args.genomebuild,
+        &args.file_date,
         &args.case_uuid,
         worker_version(),
     )
@@ -494,6 +499,7 @@ mod test {
 
         let args_common = Default::default();
         let args = super::Args {
+            file_date: String::from("20230421"),
             case_uuid: uuid::Uuid::parse_str("00000000-0000-0000-0000-000000000000").unwrap(),
             max_var_count: None,
             path_mehari_db: "tests/seqvars/ingest/db".into(),
@@ -526,6 +532,7 @@ mod test {
             .into();
         let args_common = Default::default();
         let args = super::Args {
+            file_date: String::from("20230421"),
             case_uuid: uuid::Uuid::parse_str("00000000-0000-0000-0000-000000000000").unwrap(),
             max_var_count: None,
             path_mehari_db: "tests/seqvars/ingest/db".into(),

diff --git a/...s__ingest__header__test__build_output_header_37@example_dragen.07.021.624.3.10.4.vcf.snap b/...s__ingest__header__test__build_output_header_37@example_dragen.07.021.624.3.10.4.vcf.snap
@@ -46,6 +46,7 @@ expression: "std::fs::read_to_string(out_path_str)?"
 ##contig=<ID=X,length=155270560,assembly="GRCh37",species="Homo sapiens">
 ##contig=<ID=Y,length=59373566,assembly="GRCh37",species="Homo sapiens">
 ##contig=<ID=MT,length=16569,assembly="GRCh37",species="Homo sapiens">
+##fileDate=20230421
 ##SAMPLE=<ID=NA12878,Sex="Male",Disease="Affected">
 ##PEDIGREE=<ID=NA12878>
 ##x-varfish-case-uuid=00000000-0000-0000-0000-000000000000

diff --git a/...s__ingest__header__test__build_output_header_37@example_dragen.07.021.624.3.10.9.vcf.snap b/...s__ingest__header__test__build_output_header_37@example_dragen.07.021.624.3.10.9.vcf.snap
@@ -46,6 +46,7 @@ expression: "std::fs::read_to_string(out_path_str)?"
 ##contig=<ID=X,length=155270560,assembly="GRCh37",species="Homo sapiens">
 ##contig=<ID=Y,length=59373566,assembly="GRCh37",species="Homo sapiens">
 ##contig=<ID=MT,length=16569,assembly="GRCh37",species="Homo sapiens">
+##fileDate=20230421
 ##SAMPLE=<ID=CASE,Sex="Male",Disease="Affected">
 ##PEDIGREE=<ID=CASE>
 ##x-varfish-case-uuid=00000000-0000-0000-0000-000000000000

diff --git a/...ker__seqvars__ingest__header__test__build_output_header_37@example_gatk_hc.3.7-0.vcf.snap b/...ker__seqvars__ingest__header__test__build_output_header_37@example_gatk_hc.3.7-0.vcf.snap
@@ -46,6 +46,7 @@ expression: "std::fs::read_to_string(out_path_str)?"
 ##contig=<ID=X,length=155270560,assembly="GRCh37",species="Homo sapiens">
 ##contig=<ID=Y,length=59373566,assembly="GRCh37",species="Homo sapiens">
 ##contig=<ID=MT,length=16569,assembly="GRCh37",species="Homo sapiens">
+##fileDate=20230421
 ##SAMPLE=<ID=Case_1_father-N1-DNA1-WGS1,Sex="Male",Disease="Unaffected">
 ##SAMPLE=<ID=Case_1_index-N1-DNA1-WGS1,Sex="Female",Disease="Affected">
 ##SAMPLE=<ID=Case_1_mother-N1-DNA1-WGS1,Sex="Male",Disease="Unaffected">

diff --git a/...r__seqvars__ingest__header__test__build_output_header_37@example_gatk_hc.4.4.0.0.vcf.snap b/...r__seqvars__ingest__header__test__build_output_header_37@example_gatk_hc.4.4.0.0.vcf.snap
@@ -46,6 +46,7 @@ expression: "std::fs::read_to_string(out_path_str)?"
 ##contig=<ID=X,length=155270560,assembly="GRCh37",species="Homo sapiens">
 ##contig=<ID=Y,length=59373566,assembly="GRCh37",species="Homo sapiens">
 ##contig=<ID=MT,length=16569,assembly="GRCh37",species="Homo sapiens">
+##fileDate=20230421
 ##SAMPLE=<ID=CASE,Sex="Male",Disease="Affected">
 ##PEDIGREE=<ID=CASE>
 ##x-varfish-case-uuid=00000000-0000-0000-0000-000000000000

diff --git a/...s__ingest__header__test__build_output_header_38@example_dragen.07.021.624.3.10.4.vcf.snap b/...s__ingest__header__test__build_output_header_38@example_dragen.07.021.624.3.10.4.vcf.snap
@@ -46,6 +46,7 @@ expression: "std::fs::read_to_string(out_path_str)?"
 ##contig=<ID=chrX,length=156040895,assembly="GRCh38",species="Homo sapiens">
 ##contig=<ID=chrY,length=57227415,assembly="GRCh38",species="Homo sapiens">
 ##contig=<ID=chrM,length=16569,assembly="GRCh38",species="Homo sapiens">
+##fileDate=20230421
 ##SAMPLE=<ID=NA12878,Sex="Male",Disease="Affected">
 ##PEDIGREE=<ID=NA12878>
 ##x-varfish-case-uuid=00000000-0000-0000-0000-000000000000

diff --git a/...s__ingest__header__test__build_output_header_38@example_dragen.07.021.624.3.10.9.vcf.snap b/...s__ingest__header__test__build_output_header_38@example_dragen.07.021.624.3.10.9.vcf.snap
@@ -46,6 +46,7 @@ expression: "std::fs::read_to_string(out_path_str)?"
 ##contig=<ID=chrX,length=156040895,assembly="GRCh38",species="Homo sapiens">
 ##contig=<ID=chrY,length=57227415,assembly="GRCh38",species="Homo sapiens">
 ##contig=<ID=chrM,length=16569,assembly="GRCh38",species="Homo sapiens">
+##fileDate=20230421
 ##SAMPLE=<ID=CASE,Sex="Male",Disease="Affected">
 ##PEDIGREE=<ID=CASE>
 ##x-varfish-case-uuid=00000000-0000-0000-0000-000000000000

diff --git a/...ker__seqvars__ingest__header__test__build_output_header_38@example_gatk_hc.3.7-0.vcf.snap b/...ker__seqvars__ingest__header__test__build_output_header_38@example_gatk_hc.3.7-0.vcf.snap
@@ -46,6 +46,7 @@ expression: "std::fs::read_to_string(out_path_str)?"
 ##contig=<ID=chrX,length=156040895,assembly="GRCh38",species="Homo sapiens">
 ##contig=<ID=chrY,length=57227415,assembly="GRCh38",species="Homo sapiens">
 ##contig=<ID=chrM,length=16569,assembly="GRCh38",species="Homo sapiens">
+##fileDate=20230421
 ##SAMPLE=<ID=Case_1_father-N1-DNA1-WGS1,Sex="Male",Disease="Unaffected">
 ##SAMPLE=<ID=Case_1_index-N1-DNA1-WGS1,Sex="Female",Disease="Affected">
 ##SAMPLE=<ID=Case_1_mother-N1-DNA1-WGS1,Sex="Male",Disease="Unaffected">

diff --git a/...r__seqvars__ingest__header__test__build_output_header_38@example_gatk_hc.4.4.0.0.vcf.snap b/...r__seqvars__ingest__header__test__build_output_header_38@example_gatk_hc.4.4.0.0.vcf.snap
@@ -46,6 +46,7 @@ expression: "std::fs::read_to_string(out_path_str)?"
 ##contig=<ID=chrX,length=156040895,assembly="GRCh38",species="Homo sapiens">
 ##contig=<ID=chrY,length=57227415,assembly="GRCh38",species="Homo sapiens">
 ##contig=<ID=chrM,length=16569,assembly="GRCh38",species="Homo sapiens">
+##fileDate=20230421
 ##SAMPLE=<ID=CASE,Sex="Male",Disease="Affected">
 ##PEDIGREE=<ID=CASE>
 ##x-varfish-case-uuid=00000000-0000-0000-0000-000000000000

diff --git a/...pshots/varfish_server_worker__seqvars__ingest__test__result_snapshot_test@Case_1_vcf.snap b/...pshots/varfish_server_worker__seqvars__ingest__test__result_snapshot_test@Case_1_vcf.snap
@@ -46,6 +46,7 @@ expression: "std::fs::read_to_string(&args.path_out)?"
 ##contig=<ID=X,length=155270560,assembly="GRCh37",species="Homo sapiens">
 ##contig=<ID=Y,length=59373566,assembly="GRCh37",species="Homo sapiens">
 ##contig=<ID=MT,length=16569,assembly="GRCh37",species="Homo sapiens">
+##fileDate=20230421
 ##SAMPLE=<ID=Case_1_father-N1-DNA1-WGS1,Sex="Male",Disease="Unaffected">
 ##SAMPLE=<ID=Case_1_index-N1-DNA1-WGS1,Sex="Female",Disease="Affected">
 ##SAMPLE=<ID=Case_1_mother-N1-DNA1-WGS1,Sex="Female",Disease="Unaffected">

diff --git a/...arfish_server_worker__seqvars__ingest__test__result_snapshot_test@NA12878_dragen_vcf.snap b/...arfish_server_worker__seqvars__ingest__test__result_snapshot_test@NA12878_dragen_vcf.snap
@@ -46,6 +46,7 @@ expression: "std::fs::read_to_string(&args.path_out)?"
 ##contig=<ID=X,length=155270560,assembly="GRCh37",species="Homo sapiens">
 ##contig=<ID=Y,length=59373566,assembly="GRCh37",species="Homo sapiens">
 ##contig=<ID=MT,length=16569,assembly="GRCh37",species="Homo sapiens">
+##fileDate=20230421
 ##SAMPLE=<ID=NA12878,Sex="Female",Disease="Affected">
 ##PEDIGREE=<ID=NA12878>
 ##x-varfish-case-uuid=00000000-0000-0000-0000-000000000000

diff --git a/...er__seqvars__ingest__test__result_snapshot_test@example_dragen_07_021_624_3_10_4_vcf.snap b/...er__seqvars__ingest__test__result_snapshot_test@example_dragen_07_021_624_3_10_4_vcf.snap
@@ -46,6 +46,7 @@ expression: "std::fs::read_to_string(&args.path_out)?"
 ##contig=<ID=X,length=155270560,assembly="GRCh37",species="Homo sapiens">
 ##contig=<ID=Y,length=59373566,assembly="GRCh37",species="Homo sapiens">
 ##contig=<ID=MT,length=16569,assembly="GRCh37",species="Homo sapiens">
+##fileDate=20230421
 ##SAMPLE=<ID=NA12878,Sex="Male",Disease="Affected">
 ##PEDIGREE=<ID=NA12878>
 ##x-varfish-case-uuid=00000000-0000-0000-0000-000000000000

diff --git a/...er__seqvars__ingest__test__result_snapshot_test@example_dragen_07_021_624_3_10_9_vcf.snap b/...er__seqvars__ingest__test__result_snapshot_test@example_dragen_07_021_624_3_10_9_vcf.snap
@@ -46,6 +46,7 @@ expression: "std::fs::read_to_string(&args.path_out)?"
 ##contig=<ID=X,length=155270560,assembly="GRCh37",species="Homo sapiens">
 ##contig=<ID=Y,length=59373566,assembly="GRCh37",species="Homo sapiens">
 ##contig=<ID=MT,length=16569,assembly="GRCh37",species="Homo sapiens">
+##fileDate=20230421
 ##SAMPLE=<ID=CASE,Sex="Male",Disease="Affected">
 ##PEDIGREE=<ID=CASE>
 ##x-varfish-case-uuid=00000000-0000-0000-0000-000000000000

diff --git a/...server_worker__seqvars__ingest__test__result_snapshot_test@example_gatk_hc_3_7-0_vcf.snap b/...server_worker__seqvars__ingest__test__result_snapshot_test@example_gatk_hc_3_7-0_vcf.snap
@@ -46,6 +46,7 @@ expression: "std::fs::read_to_string(&args.path_out)?"
 ##contig=<ID=X,length=155270560,assembly="GRCh37",species="Homo sapiens">
 ##contig=<ID=Y,length=59373566,assembly="GRCh37",species="Homo sapiens">
 ##contig=<ID=MT,length=16569,assembly="GRCh37",species="Homo sapiens">
+##fileDate=20230421
 ##SAMPLE=<ID=Case_1_father-N1-DNA1-WGS1,Sex="Male",Disease="Unaffected">
 ##SAMPLE=<ID=Case_1_index-N1-DNA1-WGS1,Sex="Female",Disease="Affected">
 ##SAMPLE=<ID=Case_1_mother-N1-DNA1-WGS1,Sex="Male",Disease="Unaffected">

diff --git a/...rver_worker__seqvars__ingest__test__result_snapshot_test@example_gatk_hc_4_4_0_0_vcf.snap b/...rver_worker__seqvars__ingest__test__result_snapshot_test@example_gatk_hc_4_4_0_0_vcf.snap
@@ -46,6 +46,7 @@ expression: "std::fs::read_to_string(&args.path_out)?"
 ##contig=<ID=X,length=155270560,assembly="GRCh37",species="Homo sapiens">
 ##contig=<ID=Y,length=59373566,assembly="GRCh37",species="Homo sapiens">
 ##contig=<ID=MT,length=16569,assembly="GRCh37",species="Homo sapiens">
+##fileDate=20230421
 ##SAMPLE=<ID=CASE,Sex="Male",Disease="Affected">
 ##PEDIGREE=<ID=CASE>
 ##x-varfish-case-uuid=00000000-0000-0000-0000-000000000000