version 0.29.1: allow shorter final barcodes

Author: Yenaled

func_tests/runtests.sh

@@ -357,6 +357,10 @@ checkcmdoutput "$splitcode --trim-only -b AT,TT --loc-names -d 1 -p --mod-names
 checkcmdoutput "$splitcode --trim-only -b AT,TT,GGGGG -i a,a,b --loc-names -d 1 -p --mod-names $test_dir/test_bound.fq" ca97fa58494f66bd3395ac329e5a95cf
 checkcmdoutput "$splitcode --trim-only -b AT,TT,GGGGG -i a,a,a --loc-names -d 1 -p --mod-names $test_dir/test_bound.fq" ca97fa58494f66bd3395ac329e5a95cf
 
+# Shorten assigned final barcode 
+                
+checkcmdoutput "$splitcode -m /dev/null -b AT,TT -d 1 -p --mod-names --bclen=15 --bc-names $test_dir/test_bound.fq" 9e7b7980f24349c5c334e1fc02950af8
+
 # Try them with FASTA file
 
 echo ">read1

src/SplitCode.h

@@ -1,7 +1,7 @@
 #ifndef SPLITCODE_H
 #define SPLITCODE_H
 
-#define SPLITCODE_VERSION "0.29.0"
+#define SPLITCODE_VERSION "0.29.1"
 
 #include <string>
 #include <iostream>
@@ -66,6 +66,7 @@ struct SplitCode {
     summary_n_reads_filtered = 0;
     summary_n_reads_filtered_assigned = 0;
     max_seq_len = 0;
+    fake_bc_len_offset = 0;
     setNFiles(0);
     early_termination_maxFindsG = -1;
   }
@@ -74,7 +75,7 @@ struct SplitCode {
             std::string trim_5_str = "", std::string trim_3_str = "", std::string extract_str = "", bool extract_no_chain = false, std::string barcode_prefix = "",
             std::string filter_length_str = "", bool quality_trimming_5 = false, bool quality_trimming_3 = false,
             bool quality_trimming_pre = false, bool quality_trimming_naive = false, int quality_trimming_threshold = -1, bool phred64 = false,
-            bool write_tag_location_information = false, std::vector<size_t> sub_assign_vec = std::vector<size_t>(0)) {
+            bool write_tag_location_information = false, std::vector<size_t> sub_assign_vec = std::vector<size_t>(0), int fake_bc_len_override = 0) {
     init = false;
     extract_seq_names = false;
     discard_check = false;
@@ -102,6 +103,10 @@ struct SplitCode {
     this->quality_trimming_pre = quality_trimming_pre;
     this->quality_trimming_naive = quality_trimming_naive;
     this->quality_trimming_threshold = quality_trimming_threshold;
+    this->fake_bc_len_offset = 0;
+    if (fake_bc_len_override != 0) {
+      this->fake_bc_len_offset = fake_bc_len_override-((int)FAKE_BARCODE_LEN);
+    }
     this->phred64 = phred64;
     this->write_tag_location_information = write_tag_location_information;
     this->sub_assign_vec = sub_assign_vec;
@@ -3630,7 +3635,7 @@ struct SplitCode {
   }
 
   int getBarcodeLength() {
-    return FAKE_BARCODE_LEN+barcode_prefix.length();
+    return (FAKE_BARCODE_LEN+fake_bc_len_offset)+barcode_prefix.length();
   }
 
   void setNumReads(size_t num_reads, size_t max_num_reads = 0) {
@@ -3848,6 +3853,7 @@ struct SplitCode {
   int curr_barcode_mapping_i;
   int curr_umi_id_i;
   size_t max_seq_len; // Length of longest tag sequence excluding homopolymers
+  int fake_bc_len_offset;
   static const int MAX_K = 32;
   static const size_t FAKE_BARCODE_LEN = 16;
   static const char QUAL = 'K';

src/common.h

@@ -16,6 +16,7 @@ struct ProgramOptions {
   int quality_trimming_threshold;
   int64_t max_num_reads;
   int compress_level;
+  int bclen;
   bool extract_no_chain;
   bool output_fasta;
   bool no_output;
@@ -92,6 +93,7 @@ struct ProgramOptions {
     quality_trimming_threshold(-1),
     max_num_reads(0),
     compress_level(1),
+    bclen(0),
     extract_no_chain(false),
     output_fasta(false),
     no_output(false),

src/main.cpp

@@ -143,6 +143,7 @@ void usage() {
        << "-s, --summary    File where summary statistics will be written to" << endl
        << "-h, --help       Displays usage information" << endl
        << "    --assign     Assign reads to a final barcode sequence identifier based on tags present" << endl
+       << "    --bclen      The length of the final barcode sequence identifier (default: 16)" << endl
        << "    --inleaved   Specifies that input is an interleaved FASTQ file" << endl
        << "    --remultiplex  Turn on remultiplexing mode" << endl
        << "    --version    Prints version number" << endl
@@ -259,6 +260,7 @@ void ParseOptions(int argc, char **argv, ProgramOptions& opt) {
     {"sam-tags", required_argument, 0, 'M'},
     {"sub-assign", required_argument, 0, 'X'},
     {"compress", required_argument, 0, 'C'},
+    {"bclen", required_argument, 0, '9'},
     {0,0,0,0}
   };
 
@@ -441,6 +443,10 @@ void ParseOptions(int argc, char **argv, ProgramOptions& opt) {
       stringstream(optarg) >> opt.trim_3_str;
       break;
     }
+    case '9': {
+      stringstream(optarg) >> opt.bclen;
+      break;
+    }
     case 'w': {
       stringstream(optarg) >> opt.filter_length_str;
       break;
@@ -624,6 +630,16 @@ bool CheckOptions(ProgramOptions& opt, SplitCode& sc) {
            << ", but only " << n << " cores on the machine" << endl;
     }    
   }
+  if (opt.bclen != 0) {
+    if (!opt.barcode_prefix.empty()) {
+      cerr << ERROR_STR << " Cannot specify --prefix with --bclen" << endl;
+      ret = false;
+    }
+    if (opt.bclen >= 32 || opt.bclen < 2) {
+      cerr << ERROR_STR << " --bclen must have value between 2 and 32 " << endl;
+      ret = false;
+    }
+  }
   if (opt.remultiplex && opt.files.size() != 1) {
     cerr << ERROR_STR << " A single batch file must be supplied (for remultiplexing)" << endl;
     ret = false;
@@ -1297,7 +1313,7 @@ int main(int argc, char *argv[]) {
   ProgramOptions opt;
   ParseOptions(argc,argv,opt);
   SplitCode sc(opt.nfiles, opt.summary_file, opt.trim_only, opt.disable_n, opt.trim_5_str, opt.trim_3_str, opt.extract_str, opt.extract_no_chain, opt.barcode_prefix, opt.filter_length_str,
-               opt.quality_trimming_5, opt.quality_trimming_3, opt.quality_trimming_pre, opt.quality_trimming_naive, opt.quality_trimming_threshold, opt.phred64, opt.write_locations, opt.sub_assign_vec);
+               opt.quality_trimming_5, opt.quality_trimming_3, opt.quality_trimming_pre, opt.quality_trimming_naive, opt.quality_trimming_threshold, opt.phred64, opt.write_locations, opt.sub_assign_vec, opt.bclen);
   bool checkopts = CheckOptions(opt, sc);
   if (!checkopts) {
     usage();