feat: update pyft to have a writer function

py-ft/example.py

@@ -3,6 +3,7 @@
 
 bam_f = "../tests/data/center.bam"
 fiberbam = pyft.Fiberbam(bam_f)
+out_fiberbam = pyft.Fiberwriter("test.bam", bam_f)
 rgn = ["chr1", 15_323_697, 120_000_000]
 for fiber in tqdm.tqdm(fiberbam.fetch(*rgn)):
     # the number of ccs passes
@@ -16,6 +17,8 @@
     # lift reference positions to query (fiber) positions
     fiber.lift_reference_positions([0, 1, 2, 3, 4, 5, 6, 7, 8, 9])
 
+    out_fiberbam.write(fiber)
+
 
 for fiber in fiberbam.center(rgn[0], start=rgn[1], end=rgn[2], strand="-"):
     # returns the same fiber object as above; however, all the positions have been modified to be relative to the region fetched

py-ft/src/fiberdata.rs

@@ -60,6 +60,8 @@ pub struct Fiberdata {
     /// offset for centering the fiber
     #[pyo3(get, set)]
     offset: Option<i64>,
+    /// record
+    fiber: FiberseqData,
 }
 #[pymethods]
 impl Fiberdata {
@@ -121,32 +123,40 @@ impl Fiberdata {
                 strand,
             )
         };
-        let rg = if let Ok(Aux::String(f)) = fiber.record.aux(b"RG") {
+        let rg = if let std::result::Result::Ok(Aux::String(f)) = fiber.record.aux(b"RG") {
             log::trace!("{f}");
             f
         } else {
             "."
         };
 
         // fiberseq features
-        let m6a = Basemods::new(fiber.m6a.starts, fiber.m6a.reference_starts, fiber.m6a.ml);
-        let cpg = Basemods::new(fiber.cpg.starts, fiber.cpg.reference_starts, fiber.cpg.ml);
+        let m6a = Basemods::new(
+            fiber.m6a.starts.clone(),
+            fiber.m6a.reference_starts.clone(),
+            fiber.m6a.ml.clone(),
+        );
+        let cpg = Basemods::new(
+            fiber.cpg.starts.clone(),
+            fiber.cpg.reference_starts.clone(),
+            fiber.cpg.ml.clone(),
+        );
 
         let nuc = Ranges {
-            starts: fiber.nuc.starts,
-            ends: fiber.nuc.ends,
-            lengths: fiber.nuc.lengths,
-            reference_starts: fiber.nuc.reference_starts,
-            reference_ends: fiber.nuc.reference_ends,
-            reference_lengths: fiber.nuc.reference_lengths,
+            starts: fiber.nuc.starts.clone(),
+            ends: fiber.nuc.ends.clone(),
+            lengths: fiber.nuc.lengths.clone(),
+            reference_starts: fiber.nuc.reference_starts.clone(),
+            reference_ends: fiber.nuc.reference_ends.clone(),
+            reference_lengths: fiber.nuc.reference_lengths.clone(),
         };
         let msp = Ranges {
-            starts: fiber.msp.starts,
-            ends: fiber.msp.ends,
-            lengths: fiber.msp.lengths,
-            reference_starts: fiber.msp.reference_starts,
-            reference_ends: fiber.msp.reference_ends,
-            reference_lengths: fiber.msp.reference_lengths,
+            starts: fiber.msp.starts.clone(),
+            ends: fiber.msp.ends.clone(),
+            lengths: fiber.msp.lengths.clone(),
+            reference_starts: fiber.msp.reference_starts.clone(),
+            reference_ends: fiber.msp.reference_ends.clone(),
+            reference_lengths: fiber.msp.reference_lengths.clone(),
         };
 
         let aligned_block_pairs = fiber.record.aligned_block_pairs().collect();
@@ -168,10 +178,39 @@ impl Fiberdata {
             nuc,
             aligned_block_pairs,
             offset,
+            fiber,
         }
     }
 }
 
+#[pyclass]
+pub struct Fiberwriter {
+    writer: bam::Writer,
+}
+
+#[pymethods]
+impl Fiberwriter {
+    #[new]
+    pub fn new(output_bam_path: &str, template_bam_path: &str) -> Self {
+        let header = bam::Header::from_template(
+            bam::Reader::from_path(template_bam_path)
+                .expect("unable to open template bam file")
+                .header(),
+        );
+        let writer = bam::Writer::from_path(output_bam_path, &header, bam::Format::Bam)
+            .expect("unable to open output bam file");
+        Self { writer }
+    }
+    /// Write a Fiberdata object to a bam file.
+    #[pyo3(signature = (fiberdata))]
+    pub fn write(&mut self, fiberdata: &Fiberdata) -> PyResult<()> {
+        self.writer
+            .write(&fiberdata.fiber.record)
+            .expect("unable to write record");
+        pyo3::prelude::PyResult::Ok(())
+    }
+}
+
 #[pyclass]
 /// Open a fiberseq bam file. Must have an index.
 pub struct Fiberbam {

py-ft/src/lib.rs

@@ -20,5 +20,6 @@ fn pyft(_py: Python, m: &PyModule) -> PyResult<()> {
     m.add_class::<fiberdata::Fiberdata>()?;
     m.add_class::<fiberdata::Basemods>()?;
     m.add_class::<fiberdata::Ranges>()?;
+    m.add_class::<fiberdata::Fiberwriter>()?;
     Ok(())
 }