feat: Add --annotation-field option to vcf-report (#211)

* feat: Add --annotation-field option to vcf-report

* Add short parameter

src/bcf/report/oncoprint.rs

@@ -35,6 +35,7 @@ pub fn oncoprint(
     max_cells: u32,
     tsv_data_path: Option<&str>,
     plot_info: Option<Vec<String>>,
+    annotation_field: &str,
 ) -> Result<()> {
     let mut data = HashMap::new();
     let mut gene_data = HashMap::new();
@@ -64,7 +65,7 @@ pub fn oncoprint(
     for (sample, path) in sample_calls.iter().sorted() {
         let bcf_reader = bcf::Reader::from_path(path)?;
         let header_records = bcf_reader.header().header_records();
-        let ann_fields: Vec<_> = get_ann_description(header_records)?;
+        let ann_fields: Vec<_> = get_ann_description(header_records, annotation_field)?;
         clin_sig_present.insert(
             sample.to_owned(),
             ann_fields.contains(&"CLIN_SIG".to_owned()),
@@ -95,7 +96,7 @@ pub fn oncoprint(
             sample_names.push(String::from_utf8(s.to_owned())?);
         }
         let header_records = header.header_records();
-        let ann_fields: Vec<_> = get_ann_description(header_records)?;
+        let ann_fields: Vec<_> = get_ann_description(header_records, annotation_field)?;
 
         for (i, field) in ann_fields.iter().enumerate() {
             ann_indices.insert(field, i);
@@ -129,7 +130,7 @@ pub fn oncoprint(
                 .map(|s| s.to_vec())
                 .collect_vec();
 
-            let ann = record.info(b"ANN").string()?;
+            let ann = record.info(annotation_field.as_bytes()).string()?;
             if let Some(ann) = ann {
                 for alt_allele in alt_alleles {
                     let variant = if alt_allele == b"<BND>" {

src/bcf/report/table_report/create_report_table.rs

@@ -56,14 +56,15 @@ pub(crate) fn make_table_report(
     output_path: &str,
     max_read_depth: u32,
     js_files: Vec<String>,
+    annotation_field: &str,
 ) -> Result<()> {
     // HashMap<gene: String, Vec<Report>>, Vec<ann_field_identifiers: String>
     // let mut reports = HashMap::new();
     let mut ann_indices = HashMap::new();
     let mut vcf = rust_htslib::bcf::Reader::from_path(&vcf_path).unwrap();
     let header = vcf.header().clone();
     let header_records = header.header_records();
-    let ann_field_description: Vec<_> = get_ann_description(header_records)?;
+    let ann_field_description: Vec<_> = get_ann_description(header_records, annotation_field)?;
     let samples: Vec<_> = header
         .samples()
         .iter()
@@ -186,7 +187,7 @@ pub(crate) fn make_table_report(
         let mut alterations = Vec::new();
         let mut hgvsgs = Vec::new();
 
-        if let Some(ann) = variant.info(b"ANN").string()? {
+        if let Some(ann) = variant.info(annotation_field.as_bytes()).string()? {
             for entry in ann.iter() {
                 let fields = entry.split(|c| *c == b'|').collect_vec();
 
@@ -442,10 +443,13 @@ fn escape_hgvsg(hgvsg: &str) -> String {
     hgvsg.replace(".", "_").replace(">", "_").replace(":", "_")
 }
 
-pub(crate) fn get_ann_description(header_records: Vec<HeaderRecord>) -> Result<Vec<String>> {
+pub(crate) fn get_ann_description(
+    header_records: Vec<HeaderRecord>,
+    annotation_field: &str,
+) -> Result<Vec<String>> {
     for rec in header_records {
         if let rust_htslib::bcf::HeaderRecord::Info { key: _, values } = rec {
-            if values.get("ID").unwrap() == "ANN" {
+            if values.get("ID").unwrap() == annotation_field {
                 let description = values.get("Description").unwrap();
                 let fields: Vec<_> = description.split('|').collect();
                 let mut owned_fields = Vec::new();

src/bcf/report/table_report/mod.rs

@@ -20,6 +20,7 @@ pub fn table_report(
     format_strings: Option<Vec<String>>,
     max_read_depth: u32,
     js_files: Vec<String>,
+    annotation_field: &str,
 ) -> Result<()> {
     let detail_path = output_path.to_owned() + "/details/" + sample;
     fs::create_dir(Path::new(&detail_path)).context(WriteErr::CantCreateDir {
@@ -41,5 +42,6 @@ pub fn table_report(
         output_path,
         max_read_depth,
         js_files,
+        annotation_field,
     )
 }

src/cli.rs

@@ -245,7 +245,7 @@ pub(crate) enum Command {
 
     /// Creates report from a given VCF file including a visual plot
     /// for every variant with the given BAM and FASTA file.
-    /// The VCF file has to be annotated with VEP, using the options --hgvs, --hgvsg and --vcf_info_field ANN.
+    /// The VCF file has to be annotated with VEP, using the options --hgvs and --hgvsg.
     ///
     /// Examples:
     /// With current directory as default ouput path:
@@ -306,6 +306,10 @@ pub(crate) enum Command {
         #[structopt(long, default_value = "0")]
         threads: usize,
 
+        /// Set the name of the annotation field generated by VEP.
+        #[structopt(long, short = "a", default_value = "ANN")]
+        annotation_field: String,
+
         /// Relative output path for the report files. Default value is the current directory.
         #[structopt(default_value = ".")]
         output_path: String,

src/main.rs

@@ -132,6 +132,7 @@ fn main() -> Result<()> {
             custom_js_files,
             tsv,
             threads,
+            annotation_field,
             output_path,
         } => {
             let mut sample_calls = HashMap::new();
@@ -202,6 +203,7 @@ fn main() -> Result<()> {
                     formats.clone(),
                     max_read_depth,
                     js_file_names.clone(),
+                    &annotation_field,
                 )
                 .unwrap_or_else(|e| {
                     panic!("Failed building table report for sample {}. {}", sample, e)
@@ -214,6 +216,7 @@ fn main() -> Result<()> {
                 cells,
                 tsv.as_deref(),
                 plot_info,
+                &annotation_field,
             )?
         }
         VcfSplit { input, output } => bcf::split::split(input, output.as_ref())?,