feat: Enable printing FILTER field for vcf-to-txt (#242)

Users can opt in to print the FILTER field using the `--with-filter` flag

Co-authored-by: Johannes Köster <[email protected]>

src/bcf/to_txt.rs

@@ -64,11 +64,19 @@ impl Writer {
 
 const HEADER_COMMON: &[u8] = b"VARIANT";
 
-pub fn to_txt(info_tags: &[&str], format_tags: &[&str], show_genotypes: bool) -> Result<()> {
+pub fn to_txt(
+    info_tags: &[&str],
+    format_tags: &[&str],
+    show_genotypes: bool,
+    show_filter: bool,
+) -> Result<()> {
     let mut reader = bcf::Reader::from_stdin()?;
     let mut writer = Writer::new(io::BufWriter::new(io::stdout()));
 
-    let common_n = 5 + info_tags.len();
+    let mut common_n = 5 + info_tags.len();
+    if show_filter {
+        common_n += 1
+    }
     writer.write_field(HEADER_COMMON)?;
     for _ in 1..common_n {
         writer.write_field(HEADER_COMMON)?;
@@ -88,6 +96,9 @@ pub fn to_txt(info_tags: &[&str], format_tags: &[&str], show_genotypes: bool) ->
     writer.write_field(b"REF")?;
     writer.write_field(b"ALT")?;
     writer.write_field(b"QUAL")?;
+    if show_filter {
+        writer.write_field(b"FILTER")?;
+    }
     for name in info_tags {
         writer.write_field(name.as_bytes())?;
     }
@@ -124,6 +135,23 @@ pub fn to_txt(info_tags: &[&str], format_tags: &[&str], show_genotypes: bool) ->
                 q => writer.write_float(q)?,
             }
 
+            if show_filter {
+                if rec.has_filter(".".as_bytes()) {
+                    writer.write_field(b"")?
+                } else if rec.has_filter("PASS".as_bytes()) {
+                    writer.write_field(b"PASS")?
+                } else {
+                    let mut filters = Vec::new();
+                    for (i, filter) in rec.filters().enumerate() {
+                        if i != 0 {
+                            filters.push(b';');
+                        }
+                        filters.extend_from_slice(&reader.header().id_to_name(filter));
+                    }
+                    writer.write_field(&filters)?;
+                }
+            }
+
             for name in info_tags {
                 let _name = name.as_bytes();
                 if let Ok((tag_type, tag_length)) = rec.header().info_type(_name) {

src/cli.rs

@@ -110,6 +110,10 @@ pub(crate) enum Command {
         /// Display genotypes.
         #[structopt(long, short)]
         genotypes: bool,
+
+        /// Include FILTER field.
+        #[structopt(long)]
+        with_filter: bool,
     },
 
     /// Annotate for each variant in a VCF/BCF at STDIN whether it is contained in a

src/main.rs

@@ -54,10 +54,12 @@ fn main() -> Result<()> {
             info,
             format,
             genotypes,
+            with_filter,
         } => bcf::to_txt::to_txt(
             info.iter().map(|s| s as &str).collect_vec().as_slice(),
             format.iter().map(|s| s as &str).collect_vec().as_slice(),
             genotypes,
+            with_filter,
         )?,
         VcfMatch {
             vcf,

tests/expected/variant-table-with-filter.txt

@@ -0,0 +1,6 @@
+VARIANT	VARIANT	VARIANT	VARIANT	VARIANT	VARIANT	VARIANT	VARIANT	VARIANT	S1	S1	S2	S2	S3	S3
+CHROM	POS	REF	ALT	QUAL	FILTER	T	X	SOMATIC	GT	S	GT	S	GT	S
+1	100	A	T			2		true	1/1	a	./1	bbbbbbb	.	ccccccccc
+1	200	A	T		PASS	2		true	1/1	a	./1	bbbbbbb	.	ccccccccc
+1	300	A	T		IL	2		true	1/1	a	./1	bbbbbbb	.	ccccccccc
+1	400	A	T		FA;IL	2		true	1/1	a	./1	bbbbbbb	.	ccccccccc

tests/lib.rs

@@ -109,6 +109,17 @@ fn vcf_to_txt() {
 }
 
 #[test]
+fn vcf_to_txt_with_filter() {
+    assert!(Command::new("bash")
+            .arg("-c")
+            .arg("target/debug/rbt vcf-to-txt --genotypes --fmt S --info T X SOMATIC --with-filter < tests/test-with-filter.vcf > tests/variant-table-with-filter.txt")
+            .spawn().unwrap().wait().unwrap().success());
+    test_output(
+        "tests/variant-table-with-filter.txt",
+        "tests/expected/variant-table-with-filter.txt",
+    );
+}
+
 // FIXME: can't work out how to use should_panic macro
 //#[should_panic]
 fn vcf_to_txt_input_info_as_format() {

tests/test-with-filter.vcf

@@ -0,0 +1,14 @@
+##fileformat=VCFv4.3
+##FILTER=<ID=PASS,Description="All filters passed">
+##FILTER=<ID=FA,Description="oh">
+##FILTER=<ID=IL,Description="no">
+##contig=<ID=1>
+##FORMAT=<ID=S,Number=1,Type=String,Description="Text">
+##FORMAT=<ID=GT,Number=1,Type=String,Description="Genotype">
+##INFO=<ID=T,Number=A,Type=Integer,Description="Text">
+##INFO=<ID=SOMATIC,Number=0,Type=Flag,Description="Somatic variant">
+#CHROM	POS	ID	REF	ALT	QUAL	FILTER	INFO	FORMAT	S1	S2	S3
+1	100	a	A	T	.	.	T=2;SOMATIC	S:GT	a:1/1	bbbbbbb:./1	ccccccccc:.
+1	200	a	A	T	.	PASS	T=2;SOMATIC	S:GT	a:1/1	bbbbbbb:./1	ccccccccc:.
+1	300	a	A	T	.	IL	T=2;SOMATIC	S:GT	a:1/1	bbbbbbb:./1	ccccccccc:.
+1	400	a	A	T	.	FA;IL	T=2;SOMATIC	S:GT	a:1/1	bbbbbbb:./1	ccccccccc:.