Merge branch 'hotfix/1.13.3'

CHANGES.md

@@ -1,9 +1,16 @@
 # CHANGES
 
+## 1.13.3
+
+* Add gzbuffer call after gzopen to ensure we don't hit the limit where many contigs are printed
+* Added unit test for additional methods
+* Fixes #77
+* Update license dates
+
 ## 1.13.2
 
 * Futher fixes for `sentinel` calls.
-* * Correct versin of REQUIRED_MIN_LIBZ as we are looking for greater than.
+* Correct version of REQUIRED_MIN_LIBZ as we are looking for greater than.
 
 ## 1.13.1
 

Makefile

@@ -1,4 +1,4 @@
-CAVEMAN_VERSION=1.13.2
+CAVEMAN_VERSION=1.13.3
 TEST_REF?=""
 #Compiler
 CC?=gcc

src/List.c

@@ -1,5 +1,5 @@
 /**   LICENSE
-* Copyright (c) 2014-2015 Genome Research Ltd.
+* Copyright (c) 2014-2018 Genome Research Ltd.
 *
 * Author: Cancer Genome Project [email protected]
 *

src/List.h

@@ -1,5 +1,5 @@
 /**   LICENSE
-* Copyright (c) 2014-2015 Genome Research Ltd.
+* Copyright (c) 2014-2018 Genome Research Ltd.
 *
 * Author: Cancer Genome Project [email protected]
 *

src/List_algos.c

@@ -1,5 +1,5 @@
 /**   LICENSE
-* Copyright (c) 2014-2015 Genome Research Ltd.
+* Copyright (c) 2014-2018 Genome Research Ltd.
 *
 * Author: Cancer Genome Project [email protected]
 *

src/List_algos.h

@@ -1,5 +1,5 @@
 /**   LICENSE
-* Copyright (c) 2014-2015 Genome Research Ltd.
+* Copyright (c) 2014-2018 Genome Research Ltd.
 *
 * Author: Cancer Genome Project [email protected]
 *

src/alg_bean.c

@@ -1,5 +1,5 @@
 /**   LICENSE
-* Copyright (c) 2014-2015 Genome Research Ltd.
+* Copyright (c) 2014-2018 Genome Research Ltd.
 *
 * Author: Cancer Genome Project [email protected]
 *

src/alg_bean.h

@@ -1,5 +1,5 @@
 /**   LICENSE
-* Copyright (c) 2014-2015 Genome Research Ltd.
+* Copyright (c) 2014-2018 Genome Research Ltd.
 *
 * Author: Cancer Genome Project [email protected]
 *

src/algos.c

@@ -1,5 +1,5 @@
 /**   LICENSE
-* Copyright (c) 2014-2015 Genome Research Ltd.
+* Copyright (c) 2014-2018 Genome Research Ltd.
 *
 * Author: Cancer Genome Project [email protected]
 *

src/algos.h

@@ -1,5 +1,5 @@
 /**   LICENSE
-* Copyright (c) 2014-2015 Genome Research Ltd.
+* Copyright (c) 2014-2018 Genome Research Ltd.
 *
 * Author: Cancer Genome Project [email protected]
 *

src/bam_access.h

@@ -1,5 +1,5 @@
 /**   LICENSE
-* Copyright (c) 2014-2015 Genome Research Ltd.
+* Copyright (c) 2014-2018 Genome Research Ltd.
 *
 * Author: Cancer Genome Project [email protected]
 *

src/caveman.c

@@ -1,5 +1,5 @@
 /**   LICENSE
-* Copyright (c) 2014-2015 Genome Research Ltd.
+* Copyright (c) 2014-2018 Genome Research Ltd.
 *
 * Author: Cancer Genome Project [email protected]
 *

src/cn_access.c

@@ -1,5 +1,5 @@
 /**   LICENSE
-* Copyright (c) 2014-2015 Genome Research Ltd.
+* Copyright (c) 2014-2018 Genome Research Ltd.
 *
 * Author: Cancer Genome Project [email protected]
 *

src/cn_access.h

@@ -1,5 +1,5 @@
 /**   LICENSE
-* Copyright (c) 2014-2015 Genome Research Ltd.
+* Copyright (c) 2014-2018 Genome Research Ltd.
 *
 * Author: Cancer Genome Project [email protected]
 *

src/config_file_access.c

@@ -1,5 +1,5 @@
 /**   LICENSE
-* Copyright (c) 2014-2015 Genome Research Ltd.
+* Copyright (c) 2014-2018 Genome Research Ltd.
 *
 * Author: Cancer Genome Project [email protected]
 *

src/config_file_access.h

@@ -1,5 +1,5 @@
 /**   LICENSE
-* Copyright (c) 2014-2015 Genome Research Ltd.
+* Copyright (c) 2014-2018 Genome Research Ltd.
 *
 * Author: Cancer Genome Project [email protected]
 *

src/covs_access.c

@@ -1,5 +1,5 @@
 /**   LICENSE
-* Copyright (c) 2014-2015 Genome Research Ltd.
+* Copyright (c) 2014-2018 Genome Research Ltd.
 *
 * Author: Cancer Genome Project [email protected]
 *

src/covs_access.h

@@ -1,5 +1,5 @@
 /**   LICENSE
-* Copyright (c) 2014-2015 Genome Research Ltd.
+* Copyright (c) 2014-2018 Genome Research Ltd.
 *
 * Author: Cancer Genome Project [email protected]
 *

src/estep.c

@@ -1,5 +1,5 @@
 /**   LICENSE
-* Copyright (c) 2014-2015 Genome Research Ltd.
+* Copyright (c) 2014-2018 Genome Research Ltd.
 *
 * Author: Cancer Genome Project [email protected]
 *
@@ -47,6 +47,7 @@
 #include <output.h>
 #include <fai_access.h>
 #include <algos.h>
+#include <inttypes.h>
 #include <config_file_access.h>
 
 static char tum_bam_file[512];// = NULL;
@@ -88,6 +89,8 @@ static char *species = NULL;
 static char *norm_prot = "WGS";
 static char *tum_prot = "WGS";
 static int max_copy_number = 10;
+static uint32_t default_zbuffer = 1024*1024;
+static char *contig_str = "##contig=<ID=,length=,assembly=,species=>\n";
 char *valid_protocols[3] = {"WGS","WXS","RNA"};
 
 void estep_print_usage (int exit_code){
@@ -403,9 +406,9 @@ int estep_main(int argc, char *argv[]){
 
 	//Load in probability array
 	prob_arr = covs_access_read_probs_from_file(probs_file,
-														List_count(alg->read_order),List_count(alg->strand),List_count(alg->lane),
-														List_count(alg->rd_pos),List_count(alg->map_qual),List_count(alg->base_qual),
-																												List_count(alg->ref_base),List_count(alg->call_base));
+										List_count(alg->read_order),List_count(alg->strand),List_count(alg->lane),
+										List_count(alg->rd_pos),List_count(alg->map_qual),List_count(alg->base_qual),
+										List_count(alg->ref_base),List_count(alg->call_base));
 
 	//Set the algorithm modifiers and open the bam files
 	//Set the min base qual in case it's been changed.
@@ -510,15 +513,29 @@ int estep_main(int argc, char *argv[]){
 	check(chk>0,"Error generating debug file location.");
 
 	//Open files for output
+    uint32_t no_contigs = 0;
+    uint32_t total_contigs_length = 0;
+    int res_contig_cnt = fai_access_get_count_length_all_contigs(fa_file, &no_contigs, &total_contigs_length);
+    check(res_contig_cnt==0, "Error establishing contig count and name length.");
+
+    uint64_t buf_sz = (no_contigs * (strlen(contig_str) + strlen(assembly) + strlen(species) + 20 )) + total_contigs_length;
+    if (buf_sz < default_zbuffer){
+        buf_sz = default_zbuffer;
+    }
+
 	mut_file = gzopen(mut_out,"wb1");
-	check(mut_file != 0, "Error trying to open mut file for output: %s.",mut_out);
+    int buf_res = gzbuffer(mut_file, buf_sz);
+	check(mut_file!=0, "Error trying to open mut file for output: %s.",mut_out);
+    check(buf_res!=-1, "Error setting gzbuffer for file %s size to (%"PRIu32")", mut_out, buf_sz);
 	int chk_write = output_vcf_header(mut_file, tum_bam_file, norm_bam_file, fa_file,
 																									assembly, species, norm_prot, tum_prot,
 																									norm_plat, tum_plat);
 	check(chk_write==0,"Error writing header to muts file.");
 
 	snp_file = gzopen(snp_out,"wb1");
-	check(snp_file != 0, "Error trying to open snp file for output: %s.",snp_out);
+    buf_res = gzbuffer(snp_file, buf_sz);
+	check(snp_file!=0, "Error trying to open snp file for output: %s.",snp_out);
+    check(buf_res!=-1, "Error setting gzbuffer for file %s size to (%"PRIu32")", snp_out, buf_sz);
 	chk_write = output_vcf_header(snp_file, tum_bam_file, norm_bam_file, fa_file,
 																									assembly, species, norm_prot, tum_prot,
 																									norm_plat, tum_plat);
@@ -527,7 +544,9 @@ int estep_main(int argc, char *argv[]){
 
 	if(debug == 1){
 		debug_file = gzopen(debug_out,"wb1");
+        buf_res = gzbuffer(debug_file, buf_sz);
 		check(debug_file != 0, "Error trying to open snp file for output: %s.",debug_out);
+        check(buf_res != -1, "Error setting gzbuffer for file %s size to (%"PRIu32")", debug_out, buf_sz);
 		chk_write = output_vcf_header(debug_file, tum_bam_file, norm_bam_file, fa_file,
 																									assembly, species, norm_prot, tum_prot,
 																									norm_plat, tum_plat);

src/estep.h

@@ -1,5 +1,5 @@
 /**   LICENSE
-* Copyright (c) 2014-2015 Genome Research Ltd.
+* Copyright (c) 2014-2018 Genome Research Ltd.
 *
 * Author: Cancer Genome Project [email protected]
 *

src/fai_access.c

@@ -1,5 +1,5 @@
 /**   LICENSE
-* Copyright (c) 2014-2015 Genome Research Ltd.
+* Copyright (c) 2014-2018 Genome Research Ltd.
 *
 * Author: Cancer Genome Project [email protected]
 *
@@ -45,8 +45,8 @@ int fai_access_get_name_from_index(int idx, char *index_file_name, char *chr_nam
 	check(fai != NULL,"Invalid line read\n");
 	//read each chromosome until we have reached the index.
 	int i=0;
-	char rd[200];
-	while(fgets(rd, 200, fai) != NULL){
+	char rd[1000];
+	while(fgets(rd, 1000, fai) != NULL){
 		check(rd != NULL,"Invalid line read\n");
 		i++;
 		if(i == idx){
@@ -64,6 +64,36 @@ int fai_access_get_name_from_index(int idx, char *index_file_name, char *chr_nam
 	return -1;
 }
 
+int fai_access_get_count_length_all_contigs(char *fa_loc, int *count, int *total_len){
+    char *chr_name = NULL;
+    int length = 0;
+    assert(fa_loc != NULL);
+    //Open fai file
+    chr_name = malloc(sizeof(char *));
+	FILE *fai = fopen(fa_loc, "r");
+	check(fai != NULL,"Invalid line read\n");
+	//read each chromosome
+    *count = 0;
+    *total_len = 0;
+    char rd[1000];
+	while(fgets(rd, 1000, fai) != NULL){
+		check(rd != NULL,"Invalid line read\n");
+        *count = *count+1;
+        fprintf(stderr,"********** %s\n", rd);
+		int chk = sscanf(rd,"%s\t%d\t%*d\t%*d\t%*d",chr_name,&length);
+		check(chk == 2,"Wrong number of entries (%d) found in fasta index file line %s",chk,rd);
+        *total_len += strlen(chr_name);
+	}
+    free (chr_name);
+	//close file
+	check(fclose(fai)==0,"Error closing fai file.");
+	return 0;
+error:
+	if(fai)	fclose(fai);
+    if(chr_name) free(chr_name);
+	return -1;
+}
+
 char *fai_access_get_ref_seqeuence_for_pos(char *fa_loc,char *char_nom,int start_one_based,int stop){
 	assert(char_nom !=NULL);
 	assert(fa_loc != NULL);

src/fai_access.h

@@ -1,5 +1,5 @@
 /**   LICENSE
-* Copyright (c) 2014-2015 Genome Research Ltd.
+* Copyright (c) 2014-2018 Genome Research Ltd.
 *
 * Author: Cancer Genome Project [email protected]
 *
@@ -37,6 +37,7 @@
 #include "htslib/faidx.h"
 
 int fai_access_get_name_from_index(int idx, char *index_file_name, char *chr_name, int *length);
+int fai_access_get_count_length_all_contigs(char *fa_loc, int *count, int *total_len);
 char *fai_access_get_ref_seqeuence_for_pos(char *fai_loc,char *char_nom,int start_one_based,int stop);
 
 #endif

src/file_tests.c

@@ -1,5 +1,5 @@
 /**   LICENSE
-* Copyright (c) 2014-2015 Genome Research Ltd.
+* Copyright (c) 2014-2018 Genome Research Ltd.
 *
 * Author: Cancer Genome Project [email protected]
 *

src/file_tests.h

@@ -1,5 +1,5 @@
 /**   LICENSE
-* Copyright (c) 2014-2015 Genome Research Ltd.
+* Copyright (c) 2014-2018 Genome Research Ltd.
 *
 * Author: Cancer Genome Project [email protected]
 *

src/generateCavemanVCFUnmatchedNormalPanel.c

@@ -1,5 +1,5 @@
 /**   LICENSE
-* Copyright (c) 2014-2015 Genome Research Ltd.
+* Copyright (c) 2014-2018 Genome Research Ltd.
 *
 * Author: Cancer Genome Project [email protected]
 *

src/genotype.c

@@ -1,5 +1,5 @@
 /**   LICENSE
-* Copyright (c) 2014-2015 Genome Research Ltd.
+* Copyright (c) 2014-2018 Genome Research Ltd.
 *
 * Author: Cancer Genome Project [email protected]
 *

src/genotype.h

@@ -1,5 +1,5 @@
 /**   LICENSE
-* Copyright (c) 2014-2015 Genome Research Ltd.
+* Copyright (c) 2014-2018 Genome Research Ltd.
 *
 * Author: Cancer Genome Project [email protected]
 *

src/ignore_reg_access.c

@@ -1,5 +1,5 @@
 /**   LICENSE
-* Copyright (c) 2014-2015 Genome Research Ltd.
+* Copyright (c) 2014-2018 Genome Research Ltd.
 *
 * Author: Cancer Genome Project [email protected]
 *

src/ignore_reg_access.h

@@ -1,5 +1,5 @@
 /**   LICENSE
-* Copyright (c) 2014-2015 Genome Research Ltd.
+* Copyright (c) 2014-2018 Genome Research Ltd.
 *
 * Author: Cancer Genome Project [email protected]
 *

src/merge.c

@@ -1,5 +1,5 @@
 /**   LICENSE
-* Copyright (c) 2014-2015 Genome Research Ltd.
+* Copyright (c) 2014-2018 Genome Research Ltd.
 *
 * Author: Cancer Genome Project [email protected]
 *

src/merge.h

@@ -1,5 +1,5 @@
 /**   LICENSE
-* Copyright (c) 2014-2015 Genome Research Ltd.
+* Copyright (c) 2014-2018 Genome Research Ltd.
 *
 * Author: Cancer Genome Project [email protected]
 *

src/mstep.c

@@ -1,5 +1,5 @@
 /**   LICENSE
-* Copyright (c) 2014-2015 Genome Research Ltd.
+* Copyright (c) 2014-2018 Genome Research Ltd.
 *
 * Author: Cancer Genome Project [email protected]
 *

src/mstep.h

@@ -1,5 +1,5 @@
 /**   LICENSE
-* Copyright (c) 2014-2015 Genome Research Ltd.
+* Copyright (c) 2014-2018 Genome Research Ltd.
 *
 * Author: Cancer Genome Project [email protected]
 *

src/output.c

@@ -1,5 +1,5 @@
 /**   LICENSE
-* Copyright (c) 2014-2015 Genome Research Ltd.
+* Copyright (c) 2014-2018 Genome Research Ltd.
 *
 * Author: Cancer Genome Project [email protected]
 *

src/output.h

@@ -1,5 +1,5 @@
 /**   LICENSE
-* Copyright (c) 2014-2015 Genome Research Ltd.
+* Copyright (c) 2014-2018 Genome Research Ltd.
 *
 * Author: Cancer Genome Project [email protected]
 *

src/setup.c

@@ -1,5 +1,5 @@
 /**   LICENSE
-* Copyright (c) 2014-2015 Genome Research Ltd.
+* Copyright (c) 2014-2018 Genome Research Ltd.
 *
 * Author: Cancer Genome Project [email protected]
 *

src/setup.h

@@ -1,5 +1,5 @@
 /**   LICENSE
-* Copyright (c) 2014-2015 Genome Research Ltd.
+* Copyright (c) 2014-2018 Genome Research Ltd.
 *
 * Author: Cancer Genome Project [email protected]
 *