Update tabbed sequence format, changed name and allow trimming of hea…

…der.

Makefile

@@ -6,7 +6,7 @@ CFLAGS+=-O2 -g
 CPPFLAGS+=-Wall -Wextra -Wno-unused-parameter -I$(HTSDIR)
 LIBS=-lz -lhts -lm
 
-SUBPROCESSES=view head count blockview fasta basetab qualtab lengthtab type validate find trim qualmap crisprtab
+SUBPROCESSES=view head count blockview fasta basetab qualtab lengthtab type validate find trim qualmap tabseq
 SUBPROCESS_FILES=$(addsuffix .c, $(addprefix $(SRC)/fqprocess_, $(SUBPROCESSES)))
 SUBPROCESS_OBJECTS=$(addsuffix .o, $(addprefix $(SRC)/fqprocess_, $(SUBPROCESSES)))
 

src/fqheader.h

@@ -23,7 +23,7 @@
 
 // Set the current version data:
 #define PROG_NAME "fqtools"
-#define FQTOOLS_VERSION "2.2 2019-05-08"
+#define FQTOOLS_VERSION "2.3 2019-05-08"
 
 // Define the flag type:
 typedef unsigned char fqflag;
@@ -108,6 +108,7 @@ fqstatus fqprocess_validate(int argc, const char *argv[], fqglobal options);
 fqstatus fqprocess_find(int argc, const char *argv[], fqglobal options);
 fqstatus fqprocess_trim(int argc, const char *argv[], fqglobal options);
 fqstatus fqprocess_qualmap(int argc, const char *argv[], fqglobal options);
+fqstatus fqprocess_tabseq(int argc, const char *argv[], fqglobal options);
 
 #include "fqbuffer.h"
 #include "fqfile.h"

src/fqhelp.c

@@ -71,7 +71,7 @@ void global_help(){
     printf("quality   View FASTQ file quality data\n");
     printf("header2   View FASTQ file secondary header data\n");
     printf("fasta     Convert FASTQ files to FASTA format\n");
-    printf("fasta     Convert FASTQ files to CRISPR tabbed format\n");
+    printf("tabseq    Convert FASTQ files to tabbed sequence format\n");
     printf("basetab   Tabulate FASTQ base frequencies\n");
     printf("qualtab   Tabulate FASTQ quality character frequencies\n");
     printf("lengthtab Tabulate FASTQ read lengths\n");
@@ -134,8 +134,8 @@ void fqprocess_qualmap_usage(){
     printf("usage: %s [...] qualmap [-hk] [-o OUTFILE] MAP [FILE] [FILE]\n", PROG_NAME);
 }
 
-void fqprocess_crisprtab_usage(){
-    printf("usage: %s [...] crisprtab [-h] [FILE] [FILE]\n", PROG_NAME);
+void fqprocess_tabseq_usage(){
+    printf("usage: %s [...] tabseq [-h] [FILE] [FILE]\n", PROG_NAME);
 }
 
 
@@ -380,16 +380,19 @@ void fqprocess_qualmap_help(){
     printf("    If no input file is specified, input will be read from stdin.\n");
 }
 
-void fqprocess_crisprtab_help(){
-    printf("Convert FASTQ files to CRISPR tabbed format.\n");
+void fqprocess_tabseq_help(){
+    printf("Convert FASTQ files to tabbed sequence format.\n");
     printf("\n");
-    printf("crisprtab options:\n");
+    printf("tabseq options:\n");
     printf("  -h               Show this help message and exit.\n");
+    printf("  -t               Attempt to trim read number from header.\n");    
     printf("  FILE             The fastq file(s) to view.\n");
     printf("\n");
     printf("FILE:\n");
     printf("    If no input file is specified, input will be read from stdin.\n");
     printf("\n");
-	printf("The CRISPRtab format is simply <header>\t<seq> for single-end samples,\n");
-	printf("or <header1>\t<seq1>\t<header2>\t<seq2> for paired-end samples.\n");
+	printf("The tabbed sequence format is simply <header>\\t<seq> for single-end samples,\n");
+	printf("or <header1>\\t<seq1>\\t<seq2> for paired-end samples.\n");
+    printf("if -t is specified, the read header will be trimmed of the trailing pair\n");
+    printf("numbers (\"/1\" or \"/2\"), if these are present\n");
 }

src/fqhelp.h

@@ -32,7 +32,7 @@ void fqprocess_validate_usage();
 void fqprocess_find_usage();
 void fqprocess_trim_usage();
 void fqprocess_qualmap_usage();
-void fqprocess_crisprtab_usage();
+void fqprocess_tabseq_usage();
 
 void fqprocess_view_help();
 void fqprocess_head_help();
@@ -47,4 +47,4 @@ void fqprocess_validate_help();
 void fqprocess_find_help();
 void fqprocess_trim_help();
 void fqprocess_qualmap_help();
-void fqprocess_crisprtab_help();
+void fqprocess_tabseq_help();

src/fqprocess_crisprtab.c → src/fqprocess_tabseq.c

@@ -19,75 +19,73 @@ fqfsout f_out;
 fqparser_callbacks callbacks;
 char interleaving_out;
 char paired;
-fqbuffer p1header, p2header, p1seq, p2seq;
+char trim_header;
+fqbuffer p1header, p1seq, p2seq;
 
-fqbytecount fqprocess_crisprtab_readBuffer(fqflag pair, char *b, fqbytecount b_size){
+fqbytecount fqprocess_tabseq_readBuffer(fqflag pair, char *b, fqbytecount b_size){
     return fqfile_read(&(f_in.files[pair]->file), b, b_size);
 }
 
-void fqprocess_crisprtab_headerBlock(fqflag pair, char *block, fqbytecount block_n, char final){
+void fqprocess_tabseq_headerBlock(fqflag pair, char *block, fqbytecount block_n, char final){
     if(pair == FQ_PAIR_1) fqbuffer_append(&p1header, block, block_n);
-    else fqbuffer_append(&p2header, block, block_n);
 }
 
-void fqprocess_crisprtab_sequenceBlock(fqflag pair, char *block, fqbytecount block_n, char final){
+void fqprocess_tabseq_sequenceBlock(fqflag pair, char *block, fqbytecount block_n, char final){
     if(pair == FQ_PAIR_1) fqbuffer_append(&p1seq, block, block_n);
     else fqbuffer_append(&p2seq, block, block_n);
 }
 
-void fqprocess_crisprtab_endRead(fqflag pair){
+void fqprocess_tabseq_endRead(fqflag pair){
+    int header_offset = p1header.offset;
     if((paired == 0) || (pair == FQ_PAIR_2)){
-        fwrite(p1header.data, sizeof(char), p1header.offset, stdout);
+        if(trim_header == 1){
+            if(p1header.offset > 2){
+                if((strncmp(p1header.data + p1header.offset - 2, "/1", 2) == 0) || (strncmp(p1header.data + p1header.offset - 2, "/2", 2) == 0)) header_offset = header_offset - 2;
+            }            
+        }
+        fwrite(p1header.data, sizeof(char), header_offset, stdout);
         fwrite("\t", sizeof(char), 1, stdout);
         fwrite(p1seq.data, sizeof(char), p1seq.offset, stdout);
         if(paired == 1){
-            fwrite("\t", sizeof(char), 1, stdout);
-            fwrite(p2header.data, sizeof(char), p2header.offset, stdout);
             fwrite("\t", sizeof(char), 1, stdout);
             fwrite(p2seq.data, sizeof(char), p2seq.offset, stdout);            
         }
         fwrite("\n", sizeof(char), 1, stdout);
         fflush(stdout);
         fqbuffer_reset(&p1header);
-        fqbuffer_reset(&p2header);
         fqbuffer_reset(&p1seq);
         fqbuffer_reset(&p2seq);        
     }
 }
 
-fqstatus fqprocess_crisprtab(int argc, const char *argv[], fqglobal options){
+fqstatus fqprocess_tabseq(int argc, const char *argv[], fqglobal options){
     int option;
     fqstatus result;
     char finished = 0;
+    trim_header = 0;
 
     //Initialise the output buffers:
     result = fqbuffer_init(&p1header, options.output_bufsize);
     if(result != FQ_STATUS_OK) return FQ_STATUS_FAIL;
-    result = fqbuffer_init(&p2header, options.output_bufsize);
-    if(result != FQ_STATUS_OK){
-        fqbuffer_free(&p1header);
-        return FQ_STATUS_FAIL;
-    }
     result = fqbuffer_init(&p1seq, options.output_bufsize);
     if(result != FQ_STATUS_OK){
         fqbuffer_free(&p1header);
-        fqbuffer_free(&p2header);
         return FQ_STATUS_FAIL;
     }
     result = fqbuffer_init(&p2seq, options.output_bufsize);
     if(result != FQ_STATUS_OK){
         fqbuffer_free(&p1header);
-        fqbuffer_free(&p2header);
         fqbuffer_free(&p1seq);
         return FQ_STATUS_FAIL;
     }
 
     //Parse the subcommand options:
     optind++; // Skip the subcommand argument
-    while((option = getopt(argc, (char* const*)argv, "+h")) != -1){
+    while((option = getopt(argc, (char* const*)argv, "+ht")) != -1){
         switch(option){
-            case 'h':{fqprocess_crisprtab_help(); return FQ_STATUS_OK;}
-            default:{fqprocess_crisprtab_usage(); return FQ_STATUS_FAIL;}
+            case 'h':{fqprocess_tabseq_help(); return FQ_STATUS_OK;}
+            case 't':{trim_header=1; break;}
+            default:{fqprocess_tabseq_usage(); return FQ_STATUS_FAIL;}
         }
     }
 
@@ -107,11 +105,11 @@ fqstatus fqprocess_crisprtab(int argc, const char *argv[], fqglobal options){
 
     //Set the callbacks:
     set_generic_callbacks(&callbacks);
-    callbacks.readBuffer = fqprocess_crisprtab_readBuffer;
-    callbacks.endRead = fqprocess_crisprtab_endRead;
-	callbacks.header1Block = fqprocess_crisprtab_headerBlock;
-	callbacks.header2Block = fqprocess_crisprtab_headerBlock;
-	callbacks.sequenceBlock = fqprocess_crisprtab_sequenceBlock;
+    callbacks.readBuffer = fqprocess_tabseq_readBuffer;
+    callbacks.endRead = fqprocess_tabseq_endRead;
+	callbacks.header1Block = fqprocess_tabseq_headerBlock;
+	callbacks.header2Block = fqprocess_tabseq_headerBlock;
+	callbacks.sequenceBlock = fqprocess_tabseq_sequenceBlock;
 
     // Step through the input fileset:
     do finished = fqfsin_step(&f_in);
@@ -122,7 +120,6 @@ fqstatus fqprocess_crisprtab(int argc, const char *argv[], fqglobal options){
     fqfsin_close(&f_in);
     fqfsout_close(&f_out);
     fqbuffer_free(&p1header);
-    fqbuffer_free(&p2header);
     fqbuffer_free(&p1seq);
     fqbuffer_free(&p2seq);
     return result;

src/fqtools.c

@@ -131,7 +131,7 @@ int main(int argc, const char *argv[]){
     if(strcmp(command, "find") == 0) return fqprocess_find(argc, argv, options);
     if(strcmp(command, "trim") == 0) return fqprocess_trim(argc, argv, options);
     if(strcmp(command, "qualmap") == 0) return fqprocess_qualmap(argc, argv, options);
-    if(strcmp(command, "crisprtab") == 0) return fqprocess_crisprtab(argc, argv, options);
+    if(strcmp(command, "tabseq") == 0) return fqprocess_tabseq(argc, argv, options);
 
     //If we get to here, the given command string was invalid:
     fprintf(stderr, "ERROR: unknown command: \"%s\"\n", command);