From 68550723237be3674783a86d11b3277ec09339d3 Mon Sep 17 00:00:00 2001
From: "Mitchell R. Vollger" <mrvollger@gmail.com>
Date: Fri, 27 Oct 2023 14:54:44 -0700
Subject: [PATCH] feat: add simplify options to center

---
 src/center.rs | 12 +++++++++++-
 src/cli.rs    |  3 +++
 src/main.rs   |  2 ++
 3 files changed, 16 insertions(+), 1 deletion(-)

diff --git a/src/center.rs b/src/center.rs
index 2cfd4173..954cbbd7 100644
--- a/src/center.rs
+++ b/src/center.rs
@@ -22,6 +22,7 @@ pub struct CenteredFiberData {
     center_position: CenterPosition,
     pub offset: i64,
     pub reference: bool,
+    pub simplify: bool,
 }
 
 impl CenteredFiberData {
@@ -30,6 +31,7 @@ impl CenteredFiberData {
         center_position: CenterPosition,
         dist: Option<i64>,
         reference: bool,
+        simplify: bool,
     ) -> Option<Self> {
         let (ref_offset, mol_offset) =
             CenteredFiberData::find_offsets(&fiber.record, &center_position);
@@ -43,6 +45,7 @@ impl CenteredFiberData {
             center_position,
             offset,
             reference,
+            simplify,
         })
     }
     /// find both the ref and mol offsets
@@ -75,6 +78,9 @@ impl CenteredFiberData {
 
     /// Get the sequence
     pub fn subset_sequence(&self) -> String {
+        if self.simplify {
+            return "N".to_string();
+        }
         let dist = if let Some(dist) = self.dist { dist } else { 0 };
         let seq = self.fiber.record.seq().as_bytes();
 
@@ -289,6 +295,7 @@ pub fn center(
     wide: bool,
     dist: Option<i64>,
     reference: bool,
+    simplify: bool,
     buffer: &mut Box<dyn std::io::Write>,
 ) {
     let fiber_data = FiberseqData::from_records(records, header_view, min_ml_score);
@@ -298,7 +305,8 @@ pub fn center(
     fiber_data
         .into_par_iter()
         .map(|fiber| {
-            match CenteredFiberData::new(fiber, center_position.clone(), dist, reference) {
+            match CenteredFiberData::new(fiber, center_position.clone(), dist, reference, simplify)
+            {
                 Some(centered_fiber) => {
                     if wide {
                         centered_fiber.write()
@@ -335,6 +343,7 @@ pub fn center_fiberdata(
     wide: bool,
     dist: Option<i64>,
     reference: bool,
+    simplify: bool,
 ) {
     // header needed for the contig name...
     let header = bam::Header::from_template(bam.header());
@@ -378,6 +387,7 @@ pub fn center_fiberdata(
             wide,
             dist,
             reference,
+            simplify,
             &mut buffer,
         );
         pb.inc(1);
diff --git a/src/cli.rs b/src/cli.rs
index 1b113245..e5cf4ef9 100644
--- a/src/cli.rs
+++ b/src/cli.rs
@@ -231,6 +231,9 @@ pub enum Commands {
         /// Return relative reference position instead of relative molecular position
         #[clap(short, long)]
         reference: bool,
+        /// Replace the sequence output column with just "N".
+        #[clap(short, long)]
+        simplify: bool,
     },
     /// Remove HiFi kinetics tags from the input bam file
     ClearKinetics {
diff --git a/src/main.rs b/src/main.rs
index b6adcacd..78c13378 100644
--- a/src/main.rs
+++ b/src/main.rs
@@ -106,6 +106,7 @@ pub fn main() -> Result<(), Error> {
             dist,
             wide,
             reference,
+            simplify,
         }) => {
             // read in the bam from stdin or from a file
             let mut bam = bam::IndexedReader::from_path(bam)?;
@@ -118,6 +119,7 @@ pub fn main() -> Result<(), Error> {
                 *wide,
                 *dist,
                 *reference,
+                *simplify,
             );
         }
         #[cfg(feature = "predict")]