Translate BED/GFF to nodes

Cargo.toml

@@ -22,7 +22,7 @@ include_dir = "0.7.4"
 intervaltree = "0.2.7"
 itertools = "0.13.0"
 lexical-core = "1.0.2"
-noodles = { version = "0.85.0", features = ["async", "bgzf", "core", "fasta", "gff", "vcf"] }
+noodles = { version = "0.93.0", features = ["async", "bed", "bgzf", "core", "fasta", "gff", "gtf", "vcf"] }
 petgraph = "0.6.5"
 remove_dir_all = "1.0.0"
 rusqlite = { version = "0.32.1", features = ["bundled", "array", "session"] }

fixtures/beds/simple.bed

@@ -0,0 +1,4 @@
+m123	1	10	abc123.1	0	-	1	10	0,0,0	3	102,188,129,	0,3508,4691,
+m123	5	8	xyz.1	0	-	5	8	0,0,0	1	113,	0,
+m123	10	16	xyz.2	0	+	10	16	0,0,0	2	142,326,	0,10710,
+m123	14	23	foo.1	0	+	14	23	0,0,0	2	142,326,	0,10710,

fixtures/gffs/simple.gff3

@@ -0,0 +1,8 @@
+m123	HAVANA	gene	1	20	.	-	.	ID=ENSG00000294541.1
+m123	HAVANA	transcript	1	20	.	-	.	ID=ENST00000724296.1;Parent=ENSG00000294541.1
+m123	HAVANA	exon	4	8	.	-	.	ID=exon:ENST00000724296.1:1;Parent=ENST00000724296.1
+m123	HAVANA	exon	10	14	.	-	.	ID=exon:ENST00000724296.1:2;Parent=ENST00000724296.1
+m123	HAVANA	exon	16	19	.	-	.	ID=exon:ENST00000724296.1:3;Parent=ENST00000724296.1
+m123	ENSEMBL	gene	3	15	.	-	.	ID=ENSG00000277248.1
+m123	ENSEMBL	transcript	3	15	.	-	.	ID=ENST00000615943.1;Parent=ENSG00000277248.1
+m123	ENSEMBL	exon	3	15	.	-	.	ID=exon:ENST00000615943.1:1;Parent=ENST00000615943.1

src/annotations/gff.rs

@@ -47,7 +47,7 @@ pub fn propagate_gff(
         .map(|(name, path)| (name.clone(), path.sequence(conn).len() as i64))
         .collect::<HashMap<String, i64>>();
 
-    for result in reader.records() {
+    for result in reader.record_bufs() {
         let record = result?;
         let path_name = record.reference_sequence_name().to_string();
         let annotation = Annotation {
@@ -62,7 +62,7 @@ pub fn propagate_gff(
 
         let score = record.score();
         let phase = record.phase();
-        let mut updated_record_builder = gff::Record::builder()
+        let mut updated_record_builder = gff::RecordBuf::builder()
             .set_reference_sequence_name(path_name)
             .set_source(record.source().to_string())
             .set_type(record.ty().to_string())
@@ -156,7 +156,7 @@ mod tests {
 
         for (i, result) in reader
             .expect("Could not read output file!")
-            .records()
+            .record_bufs()
             .enumerate()
         {
             let record = result.unwrap();

src/lib.rs

@@ -21,6 +21,7 @@ mod progress_bar;
 pub mod range;
 #[cfg(test)]
 pub mod test_helpers;
+pub mod translate;
 pub mod updates;
 pub mod views;