[FEATURE] add new option for reading a genome

seqan · Sep 27, 2021 · be0b88e · be0b88e
1 parent f31f991
commit be0b88e
Show file tree

Hide file tree

Showing 5 changed files with 31 additions and 9 deletions.
diff --git a/include/iGenVar.hpp b/include/iGenVar.hpp
@@ -11,6 +11,7 @@ struct cmd_arguments
 // Input:
     /* -i */ std::filesystem::path alignment_short_reads_file_path{""};
     /* -j */ std::filesystem::path alignment_long_reads_file_path{""};
+    /* -g */ std::filesystem::path genome_file_path{""};
 // Output:
     /* -o */ std::filesystem::path output_file_path{};
     /* -s */ std::string vcf_sample_name{"MYSAMPLE"};

diff --git a/src/iGenVar.cpp b/src/iGenVar.cpp
@@ -35,6 +35,11 @@ void initialize_argument_parser(seqan3::argument_parser & parser, cmd_arguments
                       "Input long read alignments in SAM or BAM format (PacBio, Oxford Nanopore, ...).",
                       seqan3::option_spec::standard,
                       seqan3::input_file_validator{{"sam", "bam"}} );
+    parser.add_option(args.genome_file_path,
+                      'g', "input_genome",
+                      "Input the reference genome in FASTA or FASTQ format.",
+                      seqan3::option_spec::standard,
+                      seqan3::input_file_validator{{"fasta", "fa", "fastq"}} );
     parser.add_option(args.output_file_path, 'o', "output",
                       "The path of the vcf output file. If no path is given, will output to standard output.",
                       seqan3::option_spec::standard,
@@ -124,18 +129,25 @@ void detect_variants_in_alignment_file(cmd_arguments const & args)
     std::vector<Junction> junctions{};
     std::map<std::string, int32_t> references_lengths{};
 
-    // short reads
-    if (args.alignment_short_reads_file_path != "")
+    if (args.genome_file_path.empty())
     {
-        seqan3::debug_stream << "Detect junctions in short reads...\n";
-        detect_junctions_in_short_reads_sam_file(junctions, references_lengths, args);
-    }
+        // short reads
+        if (!args.alignment_short_reads_file_path.empty())
+        {
+            seqan3::debug_stream << "Detect junctions in short reads...\n";
+            detect_junctions_in_short_reads_sam_file(junctions, references_lengths, args);
+        }
 
-    // long reads
-    if (args.alignment_long_reads_file_path != "")
+        // long reads
+        if (!args.alignment_long_reads_file_path.empty())
+        {
+            seqan3::debug_stream << "Detect junctions in long reads...\n";
+            detect_junctions_in_long_reads_sam_file(junctions, references_lengths, args);
+        }
+    }
+    else
     {
-        seqan3::debug_stream << "Detect junctions in long reads...\n";
-        detect_junctions_in_long_reads_sam_file(junctions, references_lengths, args);
+        seqan3::debug_stream << "Detect SNPs, insertions and deletions in short reads...\n";
     }
 
     std::sort(junctions.begin(), junctions.end());

diff --git a/test/api/detection_test.cpp b/test/api/detection_test.cpp
@@ -421,6 +421,7 @@ TEST(junction_detection, analyze_sa_tag)
     // Args
     cmd_arguments args{std::filesystem::path{}, // alignment_short_reads_file_path
                        std::filesystem::path{}, // alignment_long_reads_file_path
+                       std::filesystem::path{}, // genome_file_path
                        std::filesystem::path{}, // output_file_path
                        "MYSAMPLE",              // vcf_sample_name
                        std::filesystem::path{}, // junctions_file_path

diff --git a/test/api/input_file_test.cpp b/test/api/input_file_test.cpp
@@ -58,6 +58,7 @@ TEST(input_file, detect_junctions_in_short_read_sam_file)
 
     cmd_arguments args{default_alignment_short_reads_file_path,
                        "",
+                       empty_path, // empty genome path,
                        empty_path, // empty output path,
                        default_vcf_sample_name,
                        empty_path, // empty junctions path,
@@ -100,6 +101,7 @@ TEST(input_file, detect_junctions_in_long_reads_sam_file)
 
     cmd_arguments args{"",
                        default_alignment_long_reads_file_path,
+                       empty_path, // empty genome path,
                        empty_path, // empty output path,
                        default_vcf_sample_name,
                        empty_path, // empty junctions path,
@@ -216,6 +218,7 @@ TEST(input_file, long_read_sam_file_unsorted)
 
     cmd_arguments args{"",
                        unsorted_sam_path,
+                       empty_path, // empty genome path,
                        empty_path, // empty output path,
                        default_vcf_sample_name,
                        empty_path, // empty junctions path,
@@ -295,6 +298,7 @@ TEST(input_file, short_and_long_read_sam_file_with_different_references_lengths)
 
     cmd_arguments args{short_sam_path,
                        long_sam_path,
+                       empty_path, // empty genome path
                        empty_path, // empty output path
                        default_vcf_sample_name,
                        empty_path, // empty junctions path

diff --git a/test/cli/iGenVar_cli_test.cpp b/test/cli/iGenVar_cli_test.cpp
@@ -39,6 +39,10 @@ std::string const help_page_part_1
     "          Input long read alignments in SAM or BAM format (PacBio, Oxford\n"
     "          Nanopore, ...). Default: \"\". The input file must exist and read\n"
     "          permissions must be granted. Valid file extensions are: [sam, bam].\n"
+    "    -g, --input_genome (std::filesystem::path)\n"
+    "          Input the reference genome in FASTA or FASTQ format. Default: \"\".\n"
+    "          The input file must exist and read permissions must be granted.\n"
+    "          Valid file extensions are: [fasta, fa, fastq].\n"
     "    -o, --output (std::filesystem::path)\n"
     "          The path of the vcf output file. If no path is given, will output to\n"
     "          standard output. Default: \"\". Write permissions must be granted.\n"