Add command to update with GFA

fixtures/path-diff.gfa

@@ -0,0 +1,10 @@
+S	3.2	CGA	*
+S	4.0	AAAAAAAA	*
+S	3.0	AT	*
+S	3.5	TCGATCGATCGATCGGGAACACACAGAGA	*
+L	3.2	+	3.5	+	0M
+L	3.0	+	4.0	+	0M
+L	4.0	+	3.5	+	0M
+L	3.0	+	3.2	+	0M
+P	M-123	3.0+,3.2+,3.5+	*
+P	Child.m-123-Start-2-End-5-Node-4	3.0+,4.0+,3.5+	*

fixtures/walk-diff.gfa

@@ -0,0 +1,10 @@
+S	3.2	CGA	*
+S	3.5	TCGATCGATCGATCGGGAACACACAGAGA	*
+S	3.0	AT	*
+S	4.0	AAAAAAAA	*
+L	3.2	+	3.5	+	0M
+L	3.0	+	3.2	+	0M
+L	3.0	+	4.0	+	0M
+L	4.0	+	3.5	+	0M
+W	m123	0	123	0	100	>3.0>3.2>3.5	*
+W	M-234	0	M-234	0	100	>3.0>4.0>3.5	*

src/gfa.rs

@@ -80,3 +80,11 @@ pub fn write_links(writer: &mut BufWriter<File>, links: &Vec<Link>) {
             });
     }
 }
+
+pub fn bool_to_strand(direction: bool) -> Strand {
+    if direction {
+        Strand::Forward
+    } else {
+        Strand::Reverse
+    }
+}

src/gfa_reader.rs

@@ -158,8 +158,8 @@ pub struct Link<T: SampleType, S: Opt, U: Opt> {
 #[derive(Debug, Clone, Ord, Eq, PartialOrd, PartialEq)]
 pub struct Path<T: SampleType, S: Opt, U: Opt> {
     pub name: String,
-    pub dir: Vec<bool>,
-    pub nodes: Vec<T>,
+    pub strands: Vec<bool>,
+    pub segments: Vec<T>,
     pub overlap: U,
     pub opt: S,
 }
@@ -172,8 +172,8 @@ pub struct Walk<T: SampleType, S: Opt> {
     pub seq_id: String,
     pub seq_start: i32,
     pub seq_end: i32,
-    pub walk_dir: Vec<bool>,
-    pub walk_id: Vec<T>,
+    pub strands: Vec<bool>,
+    pub segments: Vec<T>,
     pub opt: S,
 }
 
@@ -300,8 +300,8 @@ impl<T: SampleType + Ord + Clone, S: Opt + Ord + Clone, U: Opt> Gfa<T, S, U> {
                         let overlap = split_line.next();
                         z.paths.push(Path {
                             name,
-                            dir: dirs,
-                            nodes: node_id,
+                            strands: dirs,
+                            segments: node_id,
                             overlap: U::parse1(overlap, &mut z.sequence),
                             opt: S::parse1(split_line.next(), &mut z.sequence),
                         });
@@ -320,8 +320,8 @@ impl<T: SampleType + Ord + Clone, S: Opt + Ord + Clone, U: Opt> Gfa<T, S, U> {
                             seq_id,
                             seq_start,
                             seq_end,
-                            walk_dir: w1,
-                            walk_id: w2,
+                            strands: w1,
+                            segments: w2,
                             opt: S::parse1(opt, &mut z.sequence),
                         });
                     }
@@ -381,8 +381,8 @@ impl<T: SampleType + Ord + Clone, S: Opt + Ord + Clone, U: Opt> Gfa<T, S, U> {
                     + &walk.seq_start.to_string()
                     + "-"
                     + &walk.seq_end.to_string(),
-                dir: walk.walk_dir.clone(),
-                nodes: walk.walk_id.to_vec(),
+                strands: walk.strands.clone(),
+                segments: walk.segments.to_vec(),
                 overlap: U::parse1(None, &mut self.sequence),
                 opt: walk.opt.clone(),
             });
@@ -506,7 +506,7 @@ fn walk_parser<T: SampleType>(walk: &str, s1: &mut String) -> (Vec<bool>, Vec<T>
     (dirs, node_id)
 }
 
-pub fn fill_nodes(graph: &mut Gfa<u32, (), ()>) {
+pub fn fill_segments(graph: &mut Gfa<u32, (), ()>) {
     graph.segments.sort();
 
     let mut filled_vec = Vec::new();

src/imports/gfa.rs

@@ -3,6 +3,7 @@ use rusqlite::Connection;
 use std::collections::{HashMap, HashSet};
 use std::path::Path as FilePath;
 
+use crate::gfa::bool_to_strand;
 use crate::gfa_reader::Gfa;
 use crate::models::sample::Sample;
 use crate::models::{
@@ -17,14 +18,6 @@ use crate::models::{
 };
 use crate::progress_bar::{get_handler, get_progress_bar, get_time_elapsed_bar};
 
-fn bool_to_strand(direction: bool) -> Strand {
-    if direction {
-        Strand::Forward
-    } else {
-        Strand::Reverse
-    }
-}
-
 pub fn import_gfa<'a>(
     gfa_path: &FilePath,
     collection_name: &str,
@@ -84,10 +77,10 @@ pub fn import_gfa<'a>(
         let mut source_node_id = PATH_START_NODE_ID;
         let mut source_coordinate = 0;
         let mut source_strand = Strand::Forward;
-        for (index, segment_id) in input_path.nodes.iter().enumerate() {
+        for (index, segment_id) in input_path.segments.iter().enumerate() {
             let target = sequences_by_segment_id.get(segment_id).unwrap();
             let target_node_id = *node_ids_by_segment_id.get(segment_id).unwrap();
-            let target_strand = bool_to_strand(input_path.dir[index]);
+            let target_strand = bool_to_strand(input_path.strands[index]);
             edges.insert(edge_data_from_fields(
                 source_node_id,
                 source_coordinate,
@@ -116,10 +109,10 @@ pub fn import_gfa<'a>(
         let mut source_node_id = PATH_START_NODE_ID;
         let mut source_coordinate = 0;
         let mut source_strand = Strand::Forward;
-        for (index, segment_id) in input_walk.walk_id.iter().enumerate() {
+        for (index, segment_id) in input_walk.segments.iter().enumerate() {
             let target = sequences_by_segment_id.get(segment_id).unwrap();
             let target_node_id = *node_ids_by_segment_id.get(segment_id).unwrap();
-            let target_strand = bool_to_strand(input_walk.walk_dir[index]);
+            let target_strand = bool_to_strand(input_walk.strands[index]);
             edges.insert(edge_data_from_fields(
                 source_node_id,
                 source_coordinate,
@@ -176,10 +169,10 @@ pub fn import_gfa<'a>(
         let mut source_coordinate = 0;
         let mut source_strand = Strand::Forward;
         let mut path_edge_ids = vec![];
-        for (index, segment_id) in input_path.nodes.iter().enumerate() {
+        for (index, segment_id) in input_path.segments.iter().enumerate() {
             let target = sequences_by_segment_id.get(segment_id).unwrap();
             let target_node_id = *node_ids_by_segment_id.get(segment_id).unwrap();
-            let target_strand = bool_to_strand(input_path.dir[index]);
+            let target_strand = bool_to_strand(input_path.strands[index]);
             let key = edge_data_from_fields(
                 source_node_id,
                 source_coordinate,
@@ -211,10 +204,10 @@ pub fn import_gfa<'a>(
         let mut source_coordinate = 0;
         let mut source_strand = Strand::Forward;
         let mut path_edge_ids = vec![];
-        for (index, segment_id) in input_walk.walk_id.iter().enumerate() {
+        for (index, segment_id) in input_walk.segments.iter().enumerate() {
             let target = sequences_by_segment_id.get(segment_id).unwrap();
             let target_node_id = *node_ids_by_segment_id.get(segment_id).unwrap();
-            let target_strand = bool_to_strand(input_walk.walk_dir[index]);
+            let target_strand = bool_to_strand(input_walk.strands[index]);
             let key = edge_data_from_fields(
                 source_node_id,
                 source_coordinate,

src/main.rs

@@ -22,6 +22,7 @@ use gen::patch;
 use gen::updates::fasta::update_with_fasta;
 use gen::updates::gaf::{transform_csv_to_fasta, update_with_gaf};
 use gen::updates::genbank::update_with_genbank;
+use gen::updates::gfa::update_with_gfa;
 use gen::updates::library::update_with_library;
 use gen::updates::vcf::{update_with_vcf, VcfError};
 use gen::views::patch::view_patches;
@@ -133,6 +134,9 @@ enum Commands {
         /// If a new entity is found, create it as a normal import
         #[arg(long, action, alias = "cm")]
         create_missing: bool,
+        /// A GFA file to update from
+        #[arg(long)]
+        gfa: Option<String>,
     },
     /// Update a sequence collecting using GAF results.
     #[command(name = "update-gaf", arg_required_else_help(true))]
@@ -468,6 +472,7 @@ fn main() {
             end,
             coordinate_frame,
             create_missing,
+            gfa,
         }) => {
             conn.execute("BEGIN TRANSACTION", []).unwrap();
             operation_conn.execute("BEGIN TRANSACTION", []).unwrap();
@@ -534,6 +539,18 @@ fn main() {
                     Ok(_) => {}
                     Err(e) => panic!("Failed to update. Error is: {e}"),
                 }
+            } else if let Some(gfa_path) = gfa {
+                match update_with_gfa(
+                    &conn,
+                    &operation_conn,
+                    name,
+                    sample.clone().as_deref(),
+                    &new_sample.clone().unwrap(),
+                    gfa_path,
+                ) {
+                    Ok(_) => {}
+                    Err(e) => panic!("Failed to update. Error is: {e}"),
+                }
             } else {
                 panic!("Unknown file type provided for update.");
             }

src/updates.rs

@@ -1,5 +1,6 @@
 pub mod fasta;
 pub mod gaf;
 pub mod genbank;
+pub mod gfa;
 pub mod library;
 pub mod vcf;