Implemented --avoid-pseudogene to try to avoid aligning reads to pseu…

…dogenes. This is an experimental option needing further investigation

MANUAL.markdown

@@ -894,6 +894,17 @@ HISAT2 produces an optional field, XS:A:[+-], for every spliced alignment.
 
 </td></tr>
 
+<tr><td id="hisat2-options-avoid-pseudogene">
+[`--avoid-pseudogene`]: #hisat2-options-avoid-pseudogene
+
+    --avoid-pseudogene
+
+</td><td>
+
+Try to avoid aligning reads to pseudogenes.  Note this option is experimental and needs further investigation.
+
+</td></tr>
+
 <tr><td id="hisat2-options-no-templatelen-adjustment">
 [`--no-templatelen-adjustment`]: #hisat2-options-no-templatelen-adjustment
 

doc/index.shtml

@@ -20,9 +20,10 @@
 	<!--#include virtual="sidebar.inc.shtml"-->
 	</div> <!-- End of "rightside" -->
     <div id="leftside">
-          <h2>HISAT2 2.1.0 release 6/0X/2017</h2>
+          <h2>HISAT2 2.1.0 release 6/8/2017</h2>
           <ul> 
             <li>This major version includes the first release of HISAT-genotype, which currently performs HLA typing, DNA fingerprinting analysis, CYP typing on whole genome sequencing (WGS) reads.  We plan to extend the system to analyze a whole human genome, not just a few genes.  Please refer to <a href="http://ccb.jhu.edu/hisat-genotype">the HISAT-genotype website</a> for more details.</li>
+            <li>HISAT2 can be directly compiled and executed on Windows system using Visual Studio, thanks to <a href="http://www2.warwick.ac.uk/fac/sci/systemsbiology/staff/dyer/"Nigel Dyer.</a> </li>
             <li>Implemented --new-summary to output a new style of alignment summary, which is easier to parse or machine friendly.</li>
             <li>Implemented --summary-file to output alignment summary to a file in addition to a terminal (e.g. stderr).</li>
             <li>Implemented --no-templatelen-adjustment option to disable automaitc template length adjustment for RNA-seq reads.</li>

doc/sidebar.inc.shtml

@@ -35,15 +35,15 @@
 <a href="ftp://ftp.ccb.jhu.edu/pub/infphilo/hisat2/downloads"><h2><u>Releases</u></h2></a>
 <div class="box">
  <ul>
-   <table width="100%"><tbody><tr><td>version 2.0.5</td> <td align="right">11/4/2016</td></tr>
+   <table width="100%"><tbody><tr><td>version 2.1.0</td> <td align="right">6/8/2017</td></tr>
        <tr>
-	 <td><a href="ftp://ftp.ccb.jhu.edu/pub/infphilo/hisat2/downloads/hisat2-2.0.5-source.zip" onclick="javascript: pageTracker._trackPageview('/downloads/hisat'); ">&nbsp;&nbsp;&nbsp;Source code</a></td>
+	 <td><a href="ftp://ftp.ccb.jhu.edu/pub/infphilo/hisat2/downloads/hisat2-2.1.0-source.zip" onclick="javascript: pageTracker._trackPageview('/downloads/hisat'); ">&nbsp;&nbsp;&nbsp;Source code</a></td>
        </tr>
        <tr>
-	 <td><a href="ftp://ftp.ccb.jhu.edu/pub/infphilo/hisat2/downloads/hisat2-2.0.5-Linux_x86_64.zip" onclick="javascript: pageTracker._trackPageview('/downloads/hisat'); ">&nbsp;&nbsp;&nbsp;Linux x86_64 binary</a></td>
+	 <td><a href="ftp://ftp.ccb.jhu.edu/pub/infphilo/hisat2/downloads/hisat2-2.1.0-Linux_x86_64.zip" onclick="javascript: pageTracker._trackPageview('/downloads/hisat'); ">&nbsp;&nbsp;&nbsp;Linux x86_64 binary</a></td>
        </tr>
        <tr>
-	 <td><a href="ftp://ftp.ccb.jhu.edu/pub/infphilo/hisat2/downloads/hisat2-2.0.5-OSX_x86_64.zip" onclick="javascript: pageTracker._trackPageview('/downloads/hisat'); ">&nbsp;&nbsp;&nbsp;Mac OS X x86_64 binary</a></td>
+	 <td><a href="ftp://ftp.ccb.jhu.edu/pub/infphilo/hisat2/downloads/hisat2-2.1.0-OSX_x86_64.zip" onclick="javascript: pageTracker._trackPageview('/downloads/hisat'); ">&nbsp;&nbsp;&nbsp;Mac OS X x86_64 binary</a></td>
        </tr>
        <tr>
 	 <td><a href="http://www.di.fc.ul.pt/~afalcao/hisat2_windows.html" onclick="javascript: pageTracker._trackPageview('/downloads/hisat'); ">&nbsp;&nbsp;&nbsp;Windowns binary</a></td>

hi_aligner.h

@@ -5173,7 +5173,7 @@ bool HI_Aligner<index_t, local_index_t>::reportHit(
         if(hit.splicing_dir() == SPL_UNKNOWN)
             return false;
     }
-    if(tpol.no_spliced_alignment()) {
+    if(!tpol.no_spliced_alignment() && tpol.avoid_pseudogene()) {
         if(!spliced.first) {
             assert(!spliced.second);
             const index_t max_exon_size = 10000;

hisat2.cpp

@@ -263,6 +263,7 @@ static bool pseudogeneStop;
 static bool tranMapOnly; // transcriptome mapping only
 static bool tranAssm;    // alignments selected for downstream transcript assembly such as StringTie and Cufflinks
 static string tranAssm_program;
+static bool avoid_pseudogene;
 
 #ifdef USE_SRA
 static EList<string> sra_accs;
@@ -496,6 +497,7 @@ static void resetOptions() {
     tranMapOnly = false;
     tranAssm = false;
     tranAssm_program = "";
+    avoid_pseudogene = false;
 
 #ifdef USE_SRA
     sra_accs.clear();
@@ -720,6 +722,7 @@ static struct option long_options[] = {
     {(char*)"downstream-transcriptome-assembly",   no_argument, 0,        ARG_TRANSCRIPTOME_ASSEMBLY},
     {(char*)"dta",             no_argument,        0,        ARG_TRANSCRIPTOME_ASSEMBLY},
     {(char*)"dta-cufflinks",   no_argument,        0,        ARG_TRANSCRIPTOME_ASSEMBLY_CUFFLINKS},
+    {(char*)"avoid-pseudogene",no_argument,        0,        ARG_AVOID_PSEUDOGENE},
     {(char*)"no-templatelen-adjustment",    no_argument,        0,        ARG_NO_TEMPLATELEN_ADJUSTMENT},
 #ifdef USE_SRA
     {(char*)"sra-acc",         required_argument,  0,        ARG_SRA_ACC},
@@ -871,6 +874,7 @@ static void printUsage(ostream& out) {
         << "  --tmo                              reports only those alignments within known transcriptome" << endl
         << "  --dta                              reports alignments tailored for transcript assemblers" << endl
         << "  --dta-cufflinks                    reports alignments tailored specifically for cufflinks" << endl
+        << "  --avoid-pseudogene                 tries to avoid aligning reads to pseudogenes (experimental option)" << endl
         << "  --no-templatelen-adjustment        disables template length adjustment for RNA-seq reads" << endl
         << endl
 		<< " Scoring:" << endl
@@ -1669,6 +1673,10 @@ static void parseOption(int next_option, const char *arg) {
             tranAssm_program = "cufflinks";
             break;
         }
+        case ARG_AVOID_PSEUDOGENE: {
+            avoid_pseudogene = true;
+            break;
+        }
 #ifdef USE_SRA
         case ARG_SRA_ACC: {
             tokenize(arg, ",", sra_accs); format = SRA_FASTA;
@@ -3840,7 +3848,8 @@ static void driver(
                                  no_spliced_alignment,
                                  tranMapOnly,
                                  tranAssm,
-                                 xsOnly);
+                                 xsOnly,
+                                 avoid_pseudogene);
 
         GraphPolicy gpol(max_alts_tried,
                          use_haplotype,

opts.h

@@ -172,6 +172,7 @@ enum {
     ARG_TRANSCRIPTOME_MAPPING_ONLY,
     ARG_TRANSCRIPTOME_ASSEMBLY,
     ARG_TRANSCRIPTOME_ASSEMBLY_CUFFLINKS,
+    ARG_AVOID_PSEUDOGENE,
 #ifdef USE_SRA
     ARG_SRA_ACC,
 #endif

tp.h

@@ -45,7 +45,8 @@ class TranscriptomePolicy {
                         bool no_spliced_alignment = false,
                         bool transcriptome_mapping_only = false,
                         bool transcriptome_assembly = false,
-                        bool xs_only = false)
+                        bool xs_only = false,
+                        bool avoid_pseudogene = false)
     {
         init(minIntronLen,
              maxIntronLen,
@@ -54,7 +55,8 @@ class TranscriptomePolicy {
              no_spliced_alignment,
              transcriptome_mapping_only,
              transcriptome_assembly,
-             xs_only);
+             xs_only,
+             avoid_pseudogene);
     }
 
     /**
@@ -73,7 +75,8 @@ class TranscriptomePolicy {
               bool no_spliced_alignment = false,
               bool transcriptome_mapping_only = false,
               bool transcriptome_assembly = false,
-              bool xs_only = false)
+              bool xs_only = false,
+              bool avoid_pseudogene = false)
     {
         minIntronLen_ = minIntronLen;
         maxIntronLen_ = maxIntronLen;
@@ -83,6 +86,7 @@ class TranscriptomePolicy {
         transcriptome_mapping_only_ = transcriptome_mapping_only;
         transcriptome_assembly_ = transcriptome_assembly;
         xs_only_ = xs_only;
+        avoid_pseudogene_ = avoid_pseudogene;
     }
 
     size_t minIntronLen() const { return minIntronLen_; }
@@ -93,6 +97,7 @@ class TranscriptomePolicy {
     bool transcriptome_mapping_only() const { return transcriptome_mapping_only_; }
     bool transcriptome_assembly() const { return transcriptome_assembly_; }
     bool xs_only() const { return xs_only_; }
+    bool avoid_pseudogene() const { return avoid_pseudogene_; }
 
 private:
     size_t   minIntronLen_;
@@ -107,6 +112,7 @@ class TranscriptomePolicy {
     bool transcriptome_mapping_only_;
     bool transcriptome_assembly_;
     bool xs_only_;
+    bool avoid_pseudogene_;
 };
 
 #endif /*ndef TP_H_*/