Now it is possible for dovetail overlaps even if reads are not equal …

…in length.

README.md

@@ -1,7 +1,7 @@
 # Description
 
 Program ```SeqyClean```
-Version: ```1.10.08 (2017-10-15)```
+Version: ```1.10.09 (2017-10-16)```
 
 Main purpose of this software is to pre-process NGS data in order to prepare for downstream analysis.
 

nbproject/private/configurations.xml

@@ -2,6 +2,10 @@
 <configurationDescriptor version="100">
   <logicalFolder name="root" displayName="root" projectFiles="true" kind="ROOT">
     <df root="." name="0">
+      <df name="bin">
+      </df>
+      <df name="obj">
+      </df>
       <df name="src">
         <in>Dictionary.cpp</in>
         <in>Dictionary.h</in>
@@ -59,6 +63,8 @@
         <in>util.cpp</in>
         <in>util.h</in>
       </df>
+      <df name="test">
+      </df>
       <df name="test_data">
         <df name="areads">
         </df>
@@ -90,6 +96,8 @@
         <gdb_interceptlist>
           <gdbinterceptoptions gdb_all="false" gdb_unhandled="true" gdb_unexpected="true"/>
         </gdb_interceptlist>
+        <gdb_signals>
+        </gdb_signals>
         <gdb_options>
           <DebugOptions>
           </DebugOptions>
@@ -101,18 +109,18 @@
       </nativedebugger>
       <runprofile version="9">
         <runcommandpicklist>
-          <runcommandpicklistitem>bin/./seqyclean -1 test_data/test_PE1.fastq.gz -2 test_data/test_PE2.fastq.gz -o test/test -qual 24 24 -gz</runcommandpicklistitem>
-          <runcommandpicklistitem>bin/./seqyclean -1 test_data/test_PE1.fastq.gz -2 test_data/test_PE2.fastq.gz -o test/test -qual 24 24 -gz -no_adapter_trim</runcommandpicklistitem>
-          <runcommandpicklistitem>bin/./seqyclean -1 test_data/test_PE1.fastq.gz -2 test_data/test_PE2.fastq.gz -o test/test -gz -no_adapter_trim</runcommandpicklistitem>
-          <runcommandpicklistitem>bin/./seqyclean -1 test_data/test_PE1.fastq.gz -2 test_data/test_PE2.fastq.gz -o test/test -no_adapter_trim</runcommandpicklistitem>
           <runcommandpicklistitem>bin/./seqyclean -1 test_data/test_PE1.fastq.gz -2 test_data/test_PE2.fastq.gz -o test/test -no_adapter_trim -ow</runcommandpicklistitem>
           <runcommandpicklistitem>bin/./seqyclean -1 test_data/test_PE1.fastq.gz -2 test_data/test_PE2.fastq.gz -o test/test -no_adapter_trim -ow -gz</runcommandpicklistitem>
           <runcommandpicklistitem>bin/./seqyclean -1 test_data/test_PE1.fastq.gz -2 test_data/test_PE2.fastq.gz -o test/test -overlap -ow -gz</runcommandpicklistitem>
           <runcommandpicklistitem>bin/./seqyclean -1 test_data/test_PE1.fastq.gz -2 test_data/test_PE2.fastq.gz -o test/test -ow -gz</runcommandpicklistitem>
           <runcommandpicklistitem>./bin/seqyclean -1 test_data/areads/frag_1.fastq -2 test_data/areads/frag_2.fastq -ow -o test/test -minlen 20 -minoverlap 20 -alen 30 -overlap -adp test_data/areads/adapters.fa </runcommandpicklistitem>
           <runcommandpicklistitem>./bin/seqyclean -1 test_data/areads/frag_1.fastq -2 test_data/areads/frag_2.fastq -ow -o test/test -minlen 20 -minoverlap 20 -alen 30 -overlap -adp test_data/areads/adapters.fa</runcommandpicklistitem>
+          <runcommandpicklistitem>./bin/seqyclean -1 test_data/areads/frag_1.fastq -2 test_data/areads/frag_2.fastq -ow -o test/test -minlen 20 -adp test_data/areads/adapters.fa</runcommandpicklistitem>
+          <runcommandpicklistitem>./bin/seqyclean -1 /Users/ilya/Projects/seqyclean_test/thsu/R1.fastq -2 /Users/ilya/Projects/seqyclean_test/thsu/R2.fastq -ow -o /Users/ilya/Projects/seqyclean_test/thsu/test -minlen 20</runcommandpicklistitem>
+          <runcommandpicklistitem>./bin/seqyclean -1 /Users/ilya/Projects/seqyclean_tests/thsu/R1.fastq -2 /Users/ilya/Projects/seqyclean_tests/thsu/R2.fastq -ow -o /Users/ilya/Projects/seqyclean_test/thsu/test -minlen 20</runcommandpicklistitem>
+          <runcommandpicklistitem>./bin/seqyclean -1 /Users/ilya/Projects/seqyclean_tests/thsu/R1.fastq -2 /Users/ilya/Projects/seqyclean_tests/thsu/R2.fastq -ow -o /Users/ilya/Projects/seqyclean_tests/thsu/test -minlen 20</runcommandpicklistitem>
         </runcommandpicklist>
-        <runcommand>./bin/seqyclean -1 test_data/areads/frag_1.fastq -2 test_data/areads/frag_2.fastq -ow -o test/test -minlen 20 -minoverlap 20 -alen 30 -overlap -adp test_data/areads/adapters.fa</runcommand>
+        <runcommand>./bin/seqyclean -1 /Users/ilya/Projects/seqyclean_tests/thsu/R1.fastq -2 /Users/ilya/Projects/seqyclean_tests/thsu/R2.fastq -ow -o /Users/ilya/Projects/seqyclean_tests/thsu/test -minlen 20</runcommand>
         <rundir>.</rundir>
         <buildfirst>false</buildfirst>
         <terminal-type>0</terminal-type>

nbproject/private/private.xml

@@ -11,7 +11,6 @@
     <open-files xmlns="http://www.netbeans.org/ns/projectui-open-files/2">
         <group>
             <file>file:/Users/ilya/Projects/seqyclean/src/main.cpp</file>
-            <file>file:/Users/ilya/Projects/seqyclean/src/Illumina_retro_compiler.cpp</file>
         </group>
     </open-files>
 </project-private>

src/flash.cpp

@@ -3,6 +3,7 @@
 //modify this to down-weight difference in low-quality bases
 int strdist(std::string s1, std::string s2) {
     int sum = 0;
+
     if((s1.length() == s2.length()) && (s1.length() > 0)) {
         for(unsigned int i=0; i < s1.length(); i++) {
             if(s1[i] != s2[i]) sum++;
@@ -13,6 +14,7 @@ int strdist(std::string s1, std::string s2) {
         std::cout << "Index read: " << s2 << "\n";
         sum = -1;
     }
+
     return sum;
 }
 
@@ -52,7 +54,8 @@ int find_overlap_pos_adapter(std::string seq1, std::string seq2, int adaplen) {
     //when 75% identity otherwise it checks until overlap is < minoverlap
     std::string s1 = seq1;
     std::string s2 = seq2;
-    unsigned int rlen = s1.length();
+    unsigned int rlen = min(s1.length(), s2.length());
+    //unsigned int rlen = s1.length();
 
     if(adaplen >= rlen)
     {

src/main.cpp

@@ -23,7 +23,7 @@
 
 using namespace std;
 
-std::string version = "1.10.08 (2018-10-15)";
+std::string version = "1.10.09 (2018-10-16)";
 
 /*Common parameters (default)*/
 short KMER_SIZE = 15;