rewrite of FASTQ internals

[biopieces.git] / bp_bin / blast_seq

diff --git a/bp_bin/blast_seq b/bp_bin/blast_seq
index de78c35d4a2c1419027617068e2257f1c8f8452d..3feaa655d8837aa125d316754eacc54635026178 100755 (executable)
--- a/bp_bin/blast_seq
+++ b/bp_bin/blast_seq
@@ -1,4 +1,4 @@
-#!/usr/bin/env perl -w
+#!/usr/bin/env perl
 
 # Copyright (C) 2007-2009 Martin A. Hansen.
 
@@ -26,6 +26,7 @@
 # >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
 
 
+use warnings;
 use strict;
 use Maasha::Biopieces;
 use Maasha::Common;
@@ -37,16 +38,23 @@ use Maasha::Fasta;
 
 
 my ( $options, $in, $out, $tmp_dir, $tmp_in, $tmp_out, $q_type, $s_type, $record, $entry,
-     $fh_in, $fh_out, $program );
+     $fh_in, $fh_out, $progs_ok, $program );
+
+$progs_ok = 'blastn,blastp,tblastn,blastx,tblastx';
 
 $options = Maasha::Biopieces::parse_options(
     [
-        { long => 'database', short => 'd', type => 'file',   mandatory => 'no', default => undef, allowed => undef,                                  disallowed => undef },
-        { long => 'genome',   short => 'g', type => 'genome', mandatory => 'no', default => undef, allowed => undef,                                  disallowed => undef },
-        { long => 'program',  short => 'p', type => 'string', mandatory => 'no', default => undef, allowed => 'blastn,blastp,tblastn,blastx,tblastx', disallowed => undef },
-        { long => 'e_val',    short => 'e', type => 'float',  mandatory => 'no', default => 10,    allowed => undef,                                  disallowed => undef },
-        { long => 'filter',   short => 'f', type => 'string', mandatory => 'no', default => 'no',  allowed => 'yes,no',                               disallowed => undef },
-        { long => 'cpus',     short => 'c', type => 'uint',   mandatory => 'no', default => 1,     allowed => undef,                                  disallowed => 0 },
+        { long => 'database',         short => 'd', type => 'file',   mandatory => 'no', default => undef, allowed => undef,     disallowed => undef },
+        { long => 'genome',           short => 'g', type => 'genome', mandatory => 'no', default => undef, allowed => undef,     disallowed => undef },
+        { long => 'program',          short => 'p', type => 'string', mandatory => 'no', default => undef, allowed => $progs_ok, disallowed => undef },
+        { long => 'e_val',            short => 'e', type => 'float',  mandatory => 'no', default => 10,    allowed => undef,     disallowed => undef },
+        { long => 'filter',           short => 'f', type => 'string', mandatory => 'no', default => 'no',  allowed => 'yes,no',  disallowed => undef },
+        { long => 'cpus',             short => 'c', type => 'uint',   mandatory => 'no', default => 1,     allowed => undef,     disallowed => 0 },
+        { long => 'no_gaps',          short => 'G', type => 'flag',   mandatory => 'no', default => undef, allowed => undef,     disallowed => undef },
+        { long => 'megablast',        short => 'm', type => 'flag',   mandatory => 'no', default => undef, allowed => undef,     disallowed => undef },
+        { long => 'extend_threshold', short => 'E', type => 'uint',   mandatory => 'no', default => 0,     allowed => undef,     disallowed => undef },
+        { long => 'word_size',        short => 'W', type => 'uint',   mandatory => 'no', default => 0,     allowed => undef,     disallowed => undef },
+        { long => 'single_hit',       short => 's', type => 'flag',   mandatory => 'no', default => undef, allowed => undef,     disallowed => undef },
     ]   
 );
 
@@ -88,6 +96,24 @@ if ( $options->{ 'filter' } eq 'yes' ) {
      $options->{ 'filter' } = 'F';
 }
 
+if ( $options->{ 'no_gaps' } ) {
+    $options->{ 'gapped' } = 'F';
+} else {
+    $options->{ 'gapped' } = 'T';
+}
+
+if ( $options->{ 'megablast' } ) {
+    $options->{ 'megablast' } = 'T';
+} else {
+    $options->{ 'megablast' } = 'F';
+}
+
+if ( $options->{ 'single_hit' } ) {
+    $options->{ 'single_hit' } = 1
+} else {
+    $options->{ 'single_hit' } = 0
+}
+
 if ( $options->{ 'verbose' } )
 {
     Maasha::Common::run(
@@ -96,13 +122,17 @@ if ( $options->{ 'verbose' } )
             "-p $program",
             "-e $options->{ 'e_val' }",
             "-a $options->{ 'cpus' }",
+            "-g $options->{ 'gapped' }",
+            "-n $options->{ 'megablast' }",
             "-m 8",
             "-i $tmp_in",
             "-d $options->{ 'database' }",
             "-F $options->{ 'filter' }",
+            "-P $options->{ 'single_hit' }",
+            "-W $options->{ 'word_size' }",
+            "-f $options->{ 'extend_threshold' }",
             "-o $tmp_out",
-        ),
-        1
+        )
     );
 }
 else
@@ -117,10 +147,12 @@ else
             "-i $tmp_in",
             "-d $options->{ 'database' }",
             "-F $options->{ 'filter' }",
+            "-P $options->{ 'single_hit' }",
+            "-W $options->{ 'word_size' }",
+            "-f $options->{ 'extend_threshold' }",
             "-o $tmp_out",
             "> /dev/null 2>&1"
-        ),
-        1
+        )
     );
 }
 
@@ -155,10 +187,10 @@ sub guess_database_type
     # a .phr file exists.
 
     # Returns string;
-    if ( -f $options->{ 'database' } . ".phr" ) {
-        return "protein";
+    if ( -f $options->{ 'database' } . ".phr" or -f $options->{ 'database' } . ".pal" ) {
+        return "PROTEIN";
     } else {
-        return "nucleotide";
+        return "NUCLEOTIDE";
     }
 }
 
@@ -178,13 +210,13 @@ sub guess_program
 
     my ( $program );
 
-    if ( $q_type      ne "protein" and $s_type ne "protein" ) {
+    if ( $q_type      ne "PROTEIN" and $s_type ne "PROTEIN" ) {
         $program = "blastn";
-    } elsif ( $q_type eq "protein" and $s_type eq "protein" ) {
+    } elsif ( $q_type eq "PROTEIN" and $s_type eq "PROTEIN" ) {
         $program = "blastp";
-    } elsif ( $q_type ne "protein" and $s_type eq "protein" ) {
+    } elsif ( $q_type ne "PROTEIN" and $s_type eq "PROTEIN" ) {
         $program = "blastx";
-    } elsif ( $q_type eq "protein" and $s_type ne "protein" ) {
+    } elsif ( $q_type eq "PROTEIN" and $s_type ne "PROTEIN" ) {
         $program = "tblastn";
     }