preliminary contigs + contig output saves

Author: andrewprzh

assembler/src/projects/spades/contig_output_stage.cpp

@@ -11,7 +11,7 @@
 namespace debruijn_graph {
 
 void ContigOutput::run(conj_graph_pack &gp, const char*) {
-    auto output_dir = cfg::get().output_dir;
+    auto output_dir = cfg::get().output_dir + contig_name_prefix_;
 
     OutputContigs(gp.g, output_dir + "before_rr", false);
     OutputContigsToFASTG(gp.g, output_dir + "assembly_graph", gp.components);
@@ -45,7 +45,7 @@ void ContigOutput::run(conj_graph_pack &gp, const char*) {
 
         writer.OutputPaths(gp.contig_paths, output_dir + cfg::get().co.scaffolds_name);
 
-        OutputContigsToGFA(gp.g, gp.contig_paths, cfg::get().output_dir + "assembly_graph");
+        OutputContigsToGFA(gp.g, gp.contig_paths, output_dir + "assembly_graph");
     } else {
         OutputContigs(gp.g, output_dir + "simplified_contigs", cfg::get().use_unipaths);
         OutputContigs(gp.g, output_dir + cfg::get().co.contigs_name, false);

assembler/src/projects/spades/contig_output_stage.hpp

@@ -15,15 +15,15 @@ namespace debruijn_graph {
 class ContigOutput : public spades::AssemblyStage {
 private:
     bool output_paths_;
-public:
-    ContigOutput(bool output_paths = true)
-        : AssemblyStage("Contig Output", "contig_output"), output_paths_(output_paths) { }
-
-    void load(conj_graph_pack &, const std::string &, const char *) { }
+    string contig_name_prefix_;
 
-    void save(const conj_graph_pack &, const std::string &, const char *) const { }
+public:
+    ContigOutput(bool output_paths = true, bool preliminary = false, const string& contig_name_prefix = "")
+        : AssemblyStage("Contig Output", preliminary ? "preliminary_contig_output" : "contig_output"),
+          output_paths_(output_paths), contig_name_prefix_(contig_name_prefix) { }
 
     void run(conj_graph_pack &gp, const char *);
+
 };
 
 }

assembler/src/projects/spades/launch.hpp

@@ -96,12 +96,13 @@ void assemble_genome() {
         SPAdes.add(new debruijn_graph::MismatchCorrection());
     if (cfg::get().rr_enable) {
         if (two_step_rr) {
+            string prelim_prefix = "preliminary_";
             if (cfg::get().use_intermediate_contigs)
                 SPAdes.add(new debruijn_graph::PairInfoCount(true))
                       .add(new debruijn_graph::DistanceEstimation(true))
                       .add(new debruijn_graph::RepeatResolution(true))
-                      .add(new debruijn_graph::ContigOutput())
-                      .add(new debruijn_graph::SecondPhaseSetup());
+                      .add(new debruijn_graph::ContigOutput(true, true, prelim_prefix))
+                      .add(new debruijn_graph::SecondPhaseSetup(prelim_prefix));
 
             SPAdes.add(new debruijn_graph::Simplification());
         }
@@ -119,12 +120,10 @@ void assemble_genome() {
 
         //No graph modification allowed after HybridLibrariesAligning stage!
 
-        SPAdes.add(new debruijn_graph::ContigOutput(false))
+        SPAdes.add(new debruijn_graph::ContigOutput(false, false, "pre_pe_"))
               .add(new debruijn_graph::PairInfoCount())
               .add(new debruijn_graph::DistanceEstimation())
               .add(new debruijn_graph::RepeatResolution());
-    } else {
-        SPAdes.add(new debruijn_graph::ContigOutput(false));
     }
 
     SPAdes.add(new debruijn_graph::ContigOutput());

assembler/src/projects/spades/second_phase_setup.cpp

@@ -21,7 +21,7 @@ void SecondPhaseSetup::run(conj_graph_pack &gp, const char*) {
     gp.ClearRRIndices();
     gp.ClearPaths();
 
-    std::string old_pe_contigs_filename = cfg::get().output_dir + "final_contigs.fasta";
+    std::string old_pe_contigs_filename = cfg::get().output_dir + contig_name_prefix_ + "final_contigs.fasta";
     std::string new_pe_contigs_filename = cfg::get().output_dir + "first_pe_contigs.fasta";
 
     VERIFY(path::check_existence(old_pe_contigs_filename));

assembler/src/projects/spades/second_phase_setup.hpp

@@ -12,9 +12,12 @@ namespace debruijn_graph {
 
 //todo rename
 class SecondPhaseSetup : public spades::AssemblyStage {
+private:
+    string contig_name_prefix_;
+
 public:
-    SecondPhaseSetup()
-            : AssemblyStage("Second Phase Setup", "second_phase_setup") { }
+    SecondPhaseSetup(const string& contig_name_prefix = "")
+            : AssemblyStage("Second Phase Setup", "second_phase_setup"),contig_name_prefix_(contig_name_prefix)  { }
 
     void run(conj_graph_pack &gp, const char *);
 };