test to add PE mode to cd-hit-est-2d

yesimon · Apr 27, 2017 · af7fd63 · af7fd63
1 parent e051b28
commit af7fd63
Show file tree

Hide file tree

Showing 3 changed files with 20 additions and 2 deletions.
diff --git a/cdhit-common.c++ b/cdhit-common.c++
@@ -288,6 +288,7 @@ bool Options::SetOption2D( const char *flag, const char *value )
 {
 	if( SetOptionCommon( flag, value ) ) return true;
 	if (strcmp(flag, "-i2" ) == 0) input2 = value;
+        else if (strcmp(flag, "-j2" ) == 0) input2_pe = value;
 	else if (strcmp(flag, "-s2") == 0) diff_cutoff2 = atof(value);
 	else if (strcmp(flag, "-S2") == 0) diff_cutoff_aa2 = atoi(value);
 	else return false;
@@ -376,6 +377,9 @@ void Options::Validate()
 	if( has2D ){
 		if ((diff_cutoff2<0) || (diff_cutoff2>1)) bomb_error("invalid value for -s2");
 		if (diff_cutoff_aa2<0) bomb_error("invalid value for -S2");
+          if (PE_mode) {
+            if (input2_pe.size()  == 0) bomb_error("no input file for R2 sequences for 2nd db in PE mode");
+          }
 	}
 	if (global_identity == 0) print = 1;
 	if (short_coverage < long_coverage) short_coverage = long_coverage;

diff --git a/cdhit-common.h b/cdhit-common.h
@@ -299,6 +299,7 @@ struct Options
 	string  input;
 	string  input_pe;
 	string  input2;
+	string  input2_pe;
 	string  output;
 	string  output_pe;
 

diff --git a/cdhit-est-2d.c++ b/cdhit-est-2d.c++
@@ -49,6 +49,10 @@ int main(int argc, char **argv)
 	string db_in;
 	string db_in2;
 	string db_out;
+        string db_in_pe;
+        string db_in2_pe;
+        string db_out_pe;
+
 
 	options.cluster_thd = 0.95;
 	options.NAA = 10;
@@ -67,8 +71,12 @@ int main(int argc, char **argv)
 	options.Validate();
 
 	db_in = options.input;
+        db_in_pe  = options.input_pe;
 	db_in2 = options.input2;
+        db_in2_pe = options.input2_pe;
 	db_out = options.output;
+        db_out_pe = options.output_pe;
+
 
 	InitNAA( MAX_UAA );
 	options.NAAN = NAAN_array[options.NAA];
@@ -80,10 +88,12 @@ int main(int argc, char **argv)
 		make_comp_short_word_index(options.NAA, NAAN_array, Comp_AAN_idx);
 	}
 
-	seq_db.Read( db_in.c_str(), options );
+        if ( options.PE_mode ) {seq_db.Read( db_in.c_str(), db_in_pe.c_str(), options );}
+        else                   {seq_db.Read( db_in.c_str(),                   options );}
 	cout << "total seq in db1: " << seq_db.sequences.size() << endl;
 
-	seq_db2.Read( db_in2.c_str(), options );
+        if ( options.PE_mode ) { seq_db2.Read( db_in2.c_str(), db_in2_pe.c_str(), options );}
+        else                   { seq_db2.Read( db_in2.c_str(),                    options );}
 	cout << "total seq in db2: " << seq_db2.sequences.size() << endl;
 
 	seq_db.SortDivide( options );
@@ -93,6 +103,9 @@ int main(int argc, char **argv)
 	cout << "writing non-redundant sequences from db2" << endl;
 	seq_db2.WriteClusters( db_in2.c_str(), db_out.c_str(), options );
 
+        if ( options.PE_mode ) { seq_db2.WriteClusters( db_in2.c_str(), db_in2_pe.c_str(), db_out.c_str(), db_out_pe.c_str(), options ); }
+        else                   { seq_db2.WriteClusters( db_in2.c_str(),                    db_out.c_str(),                    options ); }
+
 	seq_db2.WriteExtra2D( seq_db, options );
 	cout << "program completed !" << endl << endl;
 	end_time = current_time();