Topological correctness

fixtures/aa.gfa

@@ -0,0 +1,5 @@
+H VN:Z:1.2
+S 1 A SN:Z:123 SO:i:0 SR:i:0
+S 2 A SN:Z:123 SO:i:0 SR:i:0
+L 1 + 2 + *
+P 124 1+,2+ 0M

migrations/core/01-initial/up.sql

@@ -19,8 +19,14 @@ CREATE TABLE sequence (
  length INTEGER NOT NULL
 ) STRICT;
 
+CREATE TABLE nodes (
+ id INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL,
+ sequence_hash TEXT NOT NULL,
+ FOREIGN KEY(sequence_hash) REFERENCES sequence(hash)
+) STRICT;
+
 CREATE TABLE block_group (
- id INTEGER PRIMARY KEY NOT NULL,
+ id INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL,
  collection_name TEXT NOT NULL,
  sample_name TEXT,
  name TEXT NOT NULL,
@@ -31,7 +37,7 @@ CREATE UNIQUE INDEX block_group_uidx ON block_group(collection_name, sample_name
 CREATE UNIQUE INDEX block_group_null_sample_uidx ON block_group(collection_name, name) WHERE sample_name is null;
 
 CREATE TABLE path (
- id INTEGER PRIMARY KEY NOT NULL,
+ id INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL,
  block_group_id INTEGER NOT NULL,
  name TEXT NOT NULL,
  FOREIGN KEY(block_group_id) REFERENCES block_group(id)
@@ -40,7 +46,7 @@ CREATE UNIQUE INDEX path_uidx ON path(block_group_id, name);
 
 -- an operation from a vcf can impact multiple paths and samples, so operation is not faceted on that
 CREATE TABLE operation (
- id INTEGER PRIMARY KEY NOT NULL,
+ id INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL,
  parent_id INTEGER,
  collection_name TEXT NOT NULL,
  change_type TEXT NOT NULL,
@@ -49,36 +55,36 @@ CREATE TABLE operation (
 ) STRICT;
 
 CREATE TABLE file_addition (
- id INTEGER PRIMARY KEY NOT NULL,
+ id INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL,
  file_path TEXT NOT NULL,
  file_type TEXT NOT NULL
 ) STRICT;
 
 CREATE TABLE operation_summary (
- id INTEGER PRIMARY KEY NOT NULL,
+ id INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL,
  operation_id INTEGER NOT NULL,
  summary TEXT NOT NULL,
  FOREIGN KEY(operation_id) REFERENCES operation(id)
 ) STRICT;
 
 CREATE TABLE edges (
- id INTEGER PRIMARY KEY NOT NULL,
- source_hash TEXT NOT NULL,
+ id INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL,
+ source_node_id INTEGER,
  source_coordinate INTEGER NOT NULL,
  source_strand TEXT NOT NULL,
- target_hash TEXT NOT NULL,
+ target_node_id INTEGER,
  target_coordinate INTEGER NOT NULL,
  target_strand TEXT NOT NULL,
  chromosome_index INTEGER NOT NULL,
  phased INTEGER NOT NULL,
- FOREIGN KEY(source_hash) REFERENCES sequence(hash),
- FOREIGN KEY(target_hash) REFERENCES sequence(hash),
+ FOREIGN KEY(source_node_id) REFERENCES nodes(id),
+ FOREIGN KEY(target_node_id) REFERENCES nodes(id),
  constraint chk_phased check (phased in (0, 1))
 ) STRICT;
-CREATE UNIQUE INDEX edge_uidx ON edges(source_hash, source_coordinate, source_strand, target_hash, target_coordinate, target_strand, chromosome_index, phased);
+CREATE UNIQUE INDEX edge_uidx ON edges(source_node_id, source_coordinate, source_strand, target_node_id, target_coordinate, target_strand, chromosome_index, phased);
 
 CREATE TABLE path_edges (
- id INTEGER PRIMARY KEY NOT NULL,
+ id INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL,
  path_id INTEGER NOT NULL,
  index_in_path INTEGER NOT NULL,
  edge_id INTEGER NOT NULL,
@@ -88,19 +94,22 @@ CREATE TABLE path_edges (
 CREATE UNIQUE INDEX path_edges_uidx ON path_edges(path_id, edge_id, index_in_path);
 
 CREATE TABLE block_group_edges (
- id INTEGER PRIMARY KEY NOT NULL,
+ id INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL,
  block_group_id INTEGER NOT NULL,
  edge_id INTEGER NOT NULL,
  FOREIGN KEY(block_group_id) REFERENCES block_group(id),
  FOREIGN KEY(edge_id) REFERENCES edges(id)
 ) STRICT;
 CREATE UNIQUE INDEX block_group_edges_uidx ON block_group_edges(block_group_id, edge_id);
 
-INSERT INTO sequence (hash, sequence_type, sequence, name, file_path, "length") values ("start-node-yyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyy", "OTHER", "start-node-yyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyy", "", "", 64), ("end-node-zzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzz", "OTHER", "end-node-zzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzz", "", "", 64);
 INSERT INTO gen_metadata (db_uuid) values (lower(
  hex(randomblob(4)) || '-' || hex(randomblob(2)) || '-' || '4' ||
  substr(hex( randomblob(2)), 2) || '-' ||
  substr('AB89', 1 + (abs(random()) % 4) , 1) ||
  substr(hex(randomblob(2)), 2) || '-' ||
  hex(randomblob(6))
  ));
+INSERT INTO sequence (hash, sequence_type, sequence, name, file_path, "length") values ("start-node-yyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyy", "OTHER", "start-node-yyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyy", "", "", 64), ("end-node-zzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzz", "OTHER", "end-node-zzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzz", "", "", 64);
+INSERT INTO nodes (id, sequence_hash) values (1, "start-node-yyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyy");
+INSERT INTO nodes (id, sequence_hash) values (2, "end-node-zzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzz");
+UPDATE SQLITE_SEQUENCE SET seq = 2 WHERE name = 'nodes';

src/exports/gfa.rs

@@ -10,9 +10,9 @@ use crate::models::{
  block_group_edge::BlockGroupEdge,
  collection::Collection,
  edge::{Edge, GroupBlock},
+ node::{PATH_END_NODE_ID, PATH_START_NODE_ID},
  path::Path,
  path_edge::PathEdge,
- sequence::Sequence,
  strand::Strand,
 };
 
@@ -36,9 +36,7 @@ pub fn export_gfa(conn: &Connection, collection_name: &str, filename: &PathBuf)
 
  let mut terminal_block_ids = HashSet::new();
  for block in &blocks {
- if block.sequence_hash == Sequence::PATH_START_HASH
- || block.sequence_hash == Sequence::PATH_END_HASH
- {
+ if block.node_id == PATH_START_NODE_ID || block.node_id == PATH_END_NODE_ID {
  terminal_block_ids.insert(block.id);
  continue;
  }
@@ -109,7 +107,7 @@ fn link_line(
  target_strand: Strand,
 ) -> String {
  format!(
- "L\t{}\t{}\t{}\t{}\t*\n",
+ "L\t{}\t{}\t{}\t{}\t0M\n",
  source_index + 1,
  source_strand,
  target_index + 1,
@@ -125,21 +123,21 @@ fn link_line(
 fn nodes_for_edges(
  edge1: &Edge,
  edge2: &Edge,
- blocks_by_hash_and_start: &HashMap<(&str, i32), GroupBlock>,
- blocks_by_hash_and_end: &HashMap<(&str, i32), GroupBlock>,
+ blocks_by_node_and_start: &HashMap<(i32, i32), GroupBlock>,
+ blocks_by_node_and_end: &HashMap<(i32, i32), GroupBlock>,
 ) -> Vec<i32> {
- let mut current_block = blocks_by_hash_and_start
- .get(&(edge1.target_hash.as_str(), edge1.target_coordinate))
+ let mut current_block = blocks_by_node_and_start
+ .get(&(edge1.target_node_id, edge1.target_coordinate))
  .unwrap();
- let end_block = blocks_by_hash_and_end
- .get(&(edge2.source_hash.as_str(), edge2.source_coordinate))
+ let end_block = blocks_by_node_and_end
+ .get(&(edge2.source_node_id, edge2.source_coordinate))
  .unwrap();
  let mut node_ids = vec![];
  #[allow(clippy::while_immutable_condition)]
  while current_block.id != end_block.id {
  node_ids.push(current_block.id);
- current_block = blocks_by_hash_and_start
- .get(&(current_block.sequence_hash.as_str(), current_block.end))
+ current_block = blocks_by_node_and_start
+ .get(&(current_block.node_id, current_block.end))
  .unwrap();
  }
  node_ids.push(end_block.id);
@@ -157,34 +155,34 @@ fn write_paths(
  let edges_by_path_id =
  PathEdge::edges_for_paths(conn, paths.iter().map(|path| path.id).collect());
 
- let blocks_by_hash_and_start = blocks
+ let blocks_by_node_and_start = blocks
  .iter()
- .map(|block| ((block.sequence_hash.as_str(), block.start), block.clone()))
- .collect::<HashMap<(&str, i32), GroupBlock>>();
- let blocks_by_hash_and_end = blocks
+ .map(|block| ((block.node_id, block.start), block.clone()))
+ .collect::<HashMap<(i32, i32), GroupBlock>>();
+ let blocks_by_node_and_end = blocks
  .iter()
- .map(|block| ((block.sequence_hash.as_str(), block.end), block.clone()))
- .collect::<HashMap<(&str, i32), GroupBlock>>();
+ .map(|block| ((block.node_id, block.end), block.clone()))
+ .collect::<HashMap<(i32, i32), GroupBlock>>();
 
  for path in paths {
  let edges_for_path = edges_by_path_id.get(&path.id).unwrap();
- let mut node_ids = vec![];
+ let mut graph_node_ids = vec![];
  let mut node_strands = vec![];
  for (edge1, edge2) in edges_for_path.iter().tuple_windows() {
  let current_node_ids = nodes_for_edges(
  edge1,
  edge2,
- &blocks_by_hash_and_start,
- &blocks_by_hash_and_end,
+ &blocks_by_node_and_start,
+ &blocks_by_node_and_end,
  );
  for node_id in &current_node_ids {
- node_ids.push(*node_id);
+ graph_node_ids.push(*node_id);
  node_strands.push(edge1.target_strand);
  }
  }
 
  writer
- .write_all(&path_line(&path.name, &node_ids, &node_strands).into_bytes())
+ .write_all(&path_line(&path.name, &graph_node_ids, &node_strands).into_bytes())
  .unwrap_or_else(|_| panic!("Error writing path {} to GFA stream", path.name));
  }
 }
@@ -199,12 +197,16 @@ fn path_line(path_name: &str, node_ids: &[i32], node_strands: &[Strand]) -> Stri
  format!("P\t{}\t{}\n", path_name, nodes)
 }
 
+#[cfg(test)]
 mod tests {
  // Note this useful idiom: importing names from outer (for mod tests) scope.
  use super::*;
 
  use crate::imports::gfa::import_gfa;
- use crate::models::{block_group::BlockGroup, collection::Collection};
+ use crate::models::{
+ block_group::BlockGroup, collection::Collection, node::Node, sequence::Sequence,
+ };
+
  use crate::test_helpers::{get_connection, setup_gen_dir};
  use tempfile::tempdir;
 
@@ -232,57 +234,61 @@ mod tests {
  .sequence_type("DNA")
  .sequence("CCCC")
  .save(&conn);
+ let node1_id = Node::create(&conn, &sequence1.hash);
+ let node2_id = Node::create(&conn, &sequence2.hash);
+ let node3_id = Node::create(&conn, &sequence3.hash);
+ let node4_id = Node::create(&conn, &sequence4.hash);
 
  let edge1 = Edge::create(
  &conn,
- Sequence::PATH_START_HASH.to_string(),
+ PATH_START_NODE_ID,
  0,
  Strand::Forward,
- sequence1.hash.clone(),
+ node1_id,
  0,
  Strand::Forward,
  0,
  0,
  );
  let edge2 = Edge::create(
  &conn,
- sequence1.hash,
+ node1_id,
  4,
  Strand::Forward,
- sequence2.hash.clone(),
+ node2_id,
  0,
  Strand::Forward,
  0,
  0,
  );
  let edge3 = Edge::create(
  &conn,
- sequence2.hash,
+ node2_id,
  4,
  Strand::Forward,
- sequence3.hash.clone(),
+ node3_id,
  0,
  Strand::Forward,
  0,
  0,
  );
  let edge4 = Edge::create(
  &conn,
- sequence3.hash,
+ node3_id,
  4,
  Strand::Forward,
- sequence4.hash.clone(),
+ node4_id,
  0,
  Strand::Forward,
  0,
  0,
  );
  let edge5 = Edge::create(
  &conn,
- sequence4.hash,
+ node4_id,
  4,
  Strand::Forward,
- Sequence::PATH_END_HASH.to_string(),
+ PATH_END_NODE_ID,
  0,
  Strand::Forward,
  0,

src/imports/fasta.rs

@@ -4,8 +4,15 @@ use std::str;
 use crate::models::file_types::FileTypes;
 use crate::models::operations::{FileAddition, Operation, OperationSummary};
 use crate::models::{
- block_group::BlockGroup, block_group_edge::BlockGroupEdge, collection::Collection, edge::Edge,
- metadata, path::Path, sequence::Sequence, strand::Strand,
+ block_group::BlockGroup,
+ block_group_edge::BlockGroupEdge,
+ collection::Collection,
+ edge::Edge,
+ metadata,
+ node::{Node, PATH_END_NODE_ID, PATH_START_NODE_ID},
+ path::Path,
+ sequence::Sequence,
+ strand::Strand,
 };
 use crate::operation_management;
 use noodles::fasta;
@@ -57,24 +64,25 @@ pub fn import_fasta(
  .sequence(&sequence)
  .save(conn)
  };
+ let node_id = Node::create(conn, &seq.hash);
  let block_group = BlockGroup::create(conn, &collection.name, None, &name);
  let edge_into = Edge::create(
  conn,
- Sequence::PATH_START_HASH.to_string(),
+ PATH_START_NODE_ID,
  0,
  Strand::Forward,
- seq.hash.to_string(),
+ node_id,
  0,
  Strand::Forward,
  0,
  0,
  );
  let edge_out_of = Edge::create(
  conn,
- seq.hash.to_string(),
+ node_id,
  sequence_length,
  Strand::Forward,
- Sequence::PATH_END_HASH.to_string(),
+ PATH_END_NODE_ID,
  0,
  Strand::Forward,
  0,