Merge branch 'release/1.13.0'

CHANGES.md

@@ -1,3 +1,7 @@
+### 1.13.0
+* bam_stats - Unit tests for C code
+* bam_stats - Fix to median insert size calculation
+
 ### 1.11.0
 * bam_stats - new rna switch to give more appropriate insert size stats
 * bam_stats - more robust handling of optional RG header entries

MANIFEST

@@ -10,6 +10,15 @@ bin/xml_to_bas.pl
 c/bam_access.c
 c/bam_access.h
 c/bam_stats.c
+c/bam_stats_calcs.c
+c/bam_stats_calcs.h
+c/bam_stats_output.c
+c/bam_stats_output.h
+c/c_tests/bam_access_tests.c
+c/c_tests/bam_stats_calcs_tests.c
+c/c_tests/bam_stats_output_tests.c
+c/c_tests/minunit.h
+c/c_tests/runtests.sh
 c/dbg.h
 c/khash.h
 CHANGES.md
@@ -68,6 +77,7 @@ t/data/paired.bam
 t/data/reconcile_bas.bam
 t/data/Stats.bam
 t/data/Stats.bam.bas
+t/data/Stats.c.bam.bas
 t/data/test.bam.bas
 t/data/unpaired.bam
 t/pcap.t

c/Makefile

@@ -30,9 +30,9 @@ LFLAGS?= -L$(HTSTMP)
 LIBS =-lhts -lpthread -lz -lm
 
 # define the C source files
-SRCS = ./bam_access.c ./bam_stats.c
+SRCS = ./bam_access.c ./bam_stats_output.c ./bam_stats_calcs.c
 #Define test sources
-TEST_SRC=$(wildcard ./tests/*_tests.c)
+TEST_SRC=$(wildcard ./c_tests/*_tests.c)
 TESTS=$(patsubst %.c,%,$(TEST_SRC))
 
 # define the C object files
@@ -60,17 +60,17 @@ BAM_STATS_TARGET=../bin/bam_stats
 
 .NOTPARALLEL: test
 
-all: clean make_htslib_tmp $(BAM_STATS_TARGET) remove_htslib_tmp
+all: clean make_htslib_tmp $(BAM_STATS_TARGET) test remove_htslib_tmp
 	@echo  bam_stats compiled.
 
 $(BAM_STATS_TARGET): $(OBJS)
-	$(CC) $(CFLAGS) $(INCLUDES) -o $(BAM_STATS_TARGET) $(OBJS) $(LFLAGS) $(LIBS)
+	$(CC) $(CFLAGS) $(INCLUDES) -o $(BAM_STATS_TARGET) $(OBJS) $(LFLAGS) $(LIBS) ./bam_stats.c
 
 #Unit Tests
 test: $(BAM_STATS_TARGET)
 test: CFLAGS += $(INCLUDES) $(OBJS) $(LFLAGS) $(LIBS)
 test: $(TESTS)
-	sh ./tests/runtests.sh
+	sh ./c_tests/runtests.sh
 
 #Unit tests with coverage
 coverage: CFLAGS += --coverage

c/bam_access.c

@@ -1,6 +1,6 @@
 /*       LICENCE
 * PCAP - NGS reference implementations and helper code for the ICGC/TCGA Pan-Cancer Analysis Project
-* Copyright (C) 2014 ICGC PanCancer Project
+* Copyright (C) 2014-2016 ICGC PanCancer Project
 *
 * This program is free software; you can redistribute it and/or
 * modify it under the terms of the GNU General Public License
@@ -21,6 +21,7 @@
 #include <stdio.h>
 #include <string.h>
 #include "bam_access.h"
+#include "bam_stats_calcs.h"
 
 int get_rg_index_from_rg_store(rg_info_t **grps, char *rg, int grps_size){
   int i=0;
@@ -35,9 +36,9 @@ void parse_rg_line(char *tmp_line, rg_info_t *group) {
   char *tag = strtok(tmp_line,"\t");
   assert(strcmp(tag,"@RG")==0);
   group->id = strdup("\0");
-  group->sample = strdup("\0"); 
-  group->platform = strdup("\0"); 
-  group->platform_unit = strdup("\0"); 
+  group->sample = strdup("\0");
+  group->platform = strdup("\0");
+  group->platform_unit = strdup("\0");
   group->lib = strdup("\0");
   tag = strtok(NULL,"\t");
   while(tag != NULL){
@@ -55,7 +56,7 @@ void parse_rg_line(char *tmp_line, rg_info_t *group) {
   return;
 }
 
-rg_info_t **parse_header(bam_hdr_t *head, int *grps_size, stats_rd_t ****grp_stats){
+rg_info_t **bam_access_parse_header(bam_hdr_t *head, int *grps_size, stats_rd_t ****grp_stats){
   assert(head != NULL);
   char *line = NULL;
   rg_info_t **groups;
@@ -150,7 +151,7 @@ rg_info_t **parse_header(bam_hdr_t *head, int *grps_size, stats_rd_t ****grp_sta
   return NULL;
 }
 
-int process_reads(htsFile *input, bam_hdr_t *head, rg_info_t **grps, int grps_size, stats_rd_t ****grp_stats, int rna){
+int bam_access_process_reads(htsFile *input, bam_hdr_t *head, rg_info_t **grps, int grps_size, stats_rd_t ****grp_stats, int rna){
   assert(input != NULL);
   assert(head != NULL);
   assert(grps != NULL);
@@ -206,12 +207,12 @@ int process_reads(htsFile *input, bam_hdr_t *head, rg_info_t **grps, int grps_si
       uint32_t nm_val = bam_aux2i(nm);
       if(nm_val>0){
         (*grp_stats)[rg_index][read]->divergent += nm_val;
-        (*grp_stats)[rg_index][read]->mapped_bases += get_mapped_base_count_from_cigar(b);
+        (*grp_stats)[rg_index][read]->mapped_bases += bam_access_get_mapped_base_count_from_cigar(b);
       }else{
         (*grp_stats)[rg_index][read]->mapped_bases += (bam_endpos(b) - b->core.pos) + 1;
       }
     }else{
-      (*grp_stats)[rg_index][read]->mapped_bases += get_mapped_base_count_from_cigar(b);
+      (*grp_stats)[rg_index][read]->mapped_bases += bam_access_get_mapped_base_count_from_cigar(b);
     }
 
     // Insert size can only be calculated based on reads that are on same chr
@@ -238,7 +239,7 @@ int process_reads(htsFile *input, bam_hdr_t *head, rg_info_t **grps, int grps_si
     return -1;
 }
 
-uint64_t get_mapped_base_count_from_cigar(bam1_t *b){
+uint64_t bam_access_get_mapped_base_count_from_cigar(bam1_t *b){
 #define _cop(c) ((c)&BAM_CIGAR_MASK)
 #define _cln(c) ((c)>>BAM_CIGAR_SHIFT)
   assert(b != NULL);

c/bam_access.h

@@ -1,6 +1,6 @@
 /*       LICENCE
 * PCAP - NGS reference implementations and helper code for the ICGC/TCGA Pan-Cancer Analysis Project
-* Copyright (C) 2014 ICGC PanCancer Project
+* Copyright (C) 2014-2016 ICGC PanCancer Project
 *
 * This program is free software; you can redistribute it and/or
 * modify it under the terms of the GNU General Public License
@@ -25,8 +25,8 @@
 #include <stdlib.h>
 #include "htslib/sam.h"
 #include "dbg.h"
-
 #include "khash.h"
+
 KHASH_MAP_INIT_INT(ins,uint64_t)
 //KHASH_INIT2(ins,, khint32_t, uint64_t, 1, kh_int_hash_func, kh_int_hash_equal)
 
@@ -52,10 +52,10 @@ typedef struct{
   char *sample;
 } rg_info_t;
 
-rg_info_t **parse_header(bam_hdr_t *head, int *grps_size, stats_rd_t ****grp_stats);
+rg_info_t **bam_access_parse_header(bam_hdr_t *head, int *grps_size, stats_rd_t ****grp_stats);
 
-int process_reads(htsFile *input, bam_hdr_t *head, rg_info_t **grps, int grps_size, stats_rd_t ****grp_stats, int rna);
+int bam_access_process_reads(htsFile *input, bam_hdr_t *head, rg_info_t **grps, int grps_size, stats_rd_t ****grp_stats, int rna);
 
-uint64_t get_mapped_base_count_from_cigar(bam1_t *b);
+uint64_t bam_access_get_mapped_base_count_from_cigar(bam1_t *b);
 
 #endif