Merge branch 'release/1.14.0'

.travis.yml

@@ -11,5 +11,13 @@ install: true
 compiler:
   - gcc
 
+before_script:
+  - wget https://sourceforge.net/projects/linasm/files/linasm-1.13%28stable%29.tar.gz/download -O linasm.tar.gz 
+  - tar -xzf linasm.tar.gz
+  - pushd linasm-1.13\(stable\) && make && sudo make install prefix=/usr && popd
+  - export PATH=/usr/bin:$PATH
+  - export C_INCLUDE_PATH=/usr/include/:$C_INCLUDE_PATH
+  - export LD_LIBRARY_PATH=/usr/lib:$LD_LIBRARY_PATH
+
 script:
-  - ./setup.sh ~/wtsi-opt
+  - ./setup.sh ~/wtsi-opt

CHANGES.md

@@ -1,5 +1,12 @@
 # CHANGES
 
+## 1.14.0
+
+* Making species and assembly required commandline arguments in the estep
+* Change method by which read position index is resolved. Results in time saving.
+* linasm is now required [linasm](http://linasm.sourceforge.net/index.php)
+* New log and exp functions from [linasm](http://linasm.sourceforge.net/index.php)
+
 ## 1.13.16
 
 * Correct occasional memory blowout caused by split step not excluding an ignore region.

INSTALL.TXT

@@ -3,7 +3,9 @@ System Requirements
 
 caveman is designed to run in a compute farm/clustre environment.
 caveman depends on:
-  htslib 1.3 <https://github.com/samtools/htslib/releases/download/1.3/htslib-1.3.tar.bz2>
+  htslib >=1.3 <https://github.com/samtools/htslib/releases/download/1.3/htslib-1.3.tar.bz2>
+  zlib >=1.2.3.5 <https://www.zlib.net/>
+  linasm >=1.13 <http://linasm.sourceforge.net/>
 
 
 Compilation

Makefile

@@ -1,4 +1,4 @@
-CAVEMAN_VERSION=1.13.16
+CAVEMAN_VERSION=1.14.0
 
 TEST_REF?=""
 #Compiler
@@ -33,7 +33,7 @@ LFLAGS?=-L$(HTSTMP)
 # define any libraries to link into executable:
 #   if I want to link in libraries (libx.so or libx.a) I use the -llibname
 #   option, something like (this will link in libmylib.so and libm.so:
-LIBS =-lhts -lpthread -lz -lm -ldl
+LIBS =-lhts -lpthread -lz -lm -ldl -llinasm
 
 # define the C source files
 SRCS = ./src/file_tests.c ./src/List.c ./src/List_algos.c ./src/bam_access.c ./src/config_file_access.c ./src/fai_access.c ./src/ignore_reg_access.c ./src/alg_bean.c ./src/split_access.c ./src/covs_access.c ./src/cn_access.c ./src/genotype.c ./src/algos.c ./src/output.c ./src/setup.c ./src/split.c ./src/mstep.c ./src/merge.c ./src/estep.c

README.md

@@ -24,6 +24,7 @@ See INSTALL.TXT
 * Reference.fasta and index
 * A one based bed style format file of regions to ignore during analysis (see specified format).
 * [zlib](https://zlib.net/) >= 1.2.3.5
+* [linasm](http://linasm.sourceforge.net/index.php) >= 1.13
 
 ## Optional inputs (will result in more accurate calls)
 

setup.sh

@@ -125,6 +125,8 @@ fi
 
 export HTSLIB="$SETUP_DIR/htslib"
 
+set -e
+
 echo -n "Building CaVEMan ..."
 if [ -e "$SETUP_DIR/caveman.success" ]; then
   echo -n " previously installed ...";

src/alg_bean.c

@@ -32,6 +32,7 @@
 
 #include <assert.h>
 #include <stdlib.h>
+#include <libgen.h>
 #include <dbg.h>
 #include <alg_bean.h>
 #include <bam_access.h>
@@ -166,11 +167,11 @@ int alg_bean_get_index_for_char_arr(List *list,char *val){
 	return -1;
 }
 
-int alg_bean_get_index_for_read_pos_prop_arr(List *list,int pos,int rd_len){
-	List *lengths = List_create();
-	int i=0;
-	int last_stop = 1;
-	LIST_FOREACH(list, first, next, cur){
+List *alg_bean_get_position_list_from_read_pos_proportion_arr(List *list,int rd_len){
+    List *lengths = List_create();
+    int last_stop = 1;
+    int i=0;
+    LIST_FOREACH(list, first, next, cur){
 		float pct = *((float *)cur->value);
 		int lng = (((float)rd_len/(float)100) * pct);
 		alg_bean_intrange *range = malloc(sizeof(alg_bean_intrange));
@@ -182,11 +183,60 @@ int alg_bean_get_index_for_read_pos_prop_arr(List *list,int pos,int rd_len){
 		range->to = (range->from) + lng;
 		last_stop = range->to;
 		List_push(lengths,range);
-		i++;
+        i++;
 	}
-	int result = alg_bean_get_index_for_intrange_arr(lengths,pos);
-	List_clear_destroy(lengths);
-	return result;
+    return lengths;
+}
+
+int alg_bean_get_index_for_read_pos_prop_arr(void *_hash, int pos,int rd_len){
+    List *lengths = NULL;
+    khiter_t k;
+    khash_t(readlenpos) *h_rd_len = (khash_t(readlenpos)*) _hash;
+    k = kh_get(readlenpos, h_rd_len, rd_len);
+    lengths = kh_value(h_rd_len, k);
+    return alg_bean_get_index_for_intrange_arr(lengths, pos);
+}
+
+int alg_bean_add_read_length_arrs(alg_bean_t *bean, char* list_loc, char* contig){
+    FILE *output_rp = NULL;
+    khash_t(readlenpos) *h_rd_len;
+    char *dircpy = NULL;
+    //Create map, key is readlength, List of sections is value.
+    char *read_len_pos_arr_file = malloc(strlen(contig) + strlen(list_loc) + 6);
+    check_mem(read_len_pos_arr_file);
+    //Create filename here through name concatenation.
+    dircpy = strdup(list_loc);
+    strcpy(read_len_pos_arr_file,dirname(dircpy));
+    strcat(read_len_pos_arr_file,"/readpos.");
+    strcat(read_len_pos_arr_file,contig);
+    output_rp = fopen(read_len_pos_arr_file,"r");
+    check(output_rp != NULL, "Error opening file %s for write.",read_len_pos_arr_file);
+    free(read_len_pos_arr_file);
+
+    h_rd_len = kh_init(readlenpos);
+    char line[500];
+    int rd_len_missing = 0;
+    khiter_t k;
+    while ( fgets(line,sizeof(line),output_rp) != NULL ){
+        //75      1-2;3-38;39-56;57-66;67-76;
+        int len;
+        char list_txt [4950];
+        int chk = sscanf(line,"%d\t%s",&len,list_txt);
+        List *lenrange = alg_bean_parse_int_range(list_txt);
+        k = kh_put(readlenpos, h_rd_len, len, &rd_len_missing);
+        if(rd_len_missing){
+            kh_value(h_rd_len, k) = lenrange;
+        }
+    }
+    fclose(output_rp);
+
+    bean->read_len_pos = h_rd_len;
+    free(dircpy);
+    return 0;
+error:
+    if(dircpy) free(dircpy);
+	if(output_rp) fclose(output_rp);
+	return -1;
 }
 
 List *alg_bean_parse_str_list(char *txt){

src/alg_bean.h

@@ -35,6 +35,10 @@
 
 #include <stdio.h>
 #include <List.h>
+#include "khash.h"
+
+//New hash to store unique readlengths
+KHASH_MAP_INIT_INT(readlenpos, List *)
 
 typedef struct alg_bean_intrange{
 	int from;
@@ -58,6 +62,7 @@ typedef struct alg_bean_t{
 	 int call_base_size;
 	 List *strand;
 	 int strand_size;
+     khash_t(readlenpos) *read_len_pos;
 } alg_bean_t;
 
 int alg_bean_create_default_file(FILE *file, char *norm, char *tum);
@@ -72,7 +77,9 @@ List *alg_bean_hard_copy_char_list(List *new_list, List *old);
 int alg_bean_get_index_for_str_arr(List *list,char *value);
 int alg_bean_get_index_for_intrange_arr(List *list,int value);
 int alg_bean_get_index_for_char_arr(List *list,char *value);
-int alg_bean_get_index_for_read_pos_prop_arr(List *list,int pos,int rd_len);
+int alg_bean_get_index_for_read_pos_prop_arr(void *_hash, int pos,int rd_len);
+int alg_bean_add_read_length_arrs(alg_bean_t *bean, char* list_loc, char* contig);
+List *alg_bean_get_position_list_from_read_pos_proportion_arr(List *list,int rd_len);
 
 #define CEIL(a, b) (((a) / (b)) + (((a) % (b)) > 0 ? 1 : 0))
 #define MIN(a, b)  (((a) < (b)) ? (a) : (b))