scbi_fqbin 0.2.2 → 0.2.3
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +7 -0
- data/.DS_Store +0 -0
- data/.gitignore +14 -0
- data/Gemfile +4 -0
- data/LICENSE.txt +22 -0
- data/{README.rdoc → README.md} +0 -0
- data/Rakefile +8 -28
- data/lib/scbi_fqbin.rb +3 -5
- data/lib/scbi_fqbin/fastabin.rb +411 -0
- data/lib/scbi_fqbin/fastq_file_c.rb +373 -0
- data/lib/scbi_fqbin/fbin_file.rb +1 -1
- data/lib/scbi_fqbin/t.rb +9 -0
- data/lib/scbi_fqbin/t2.rb +12 -0
- data/lib/scbi_fqbin/version.rb +3 -0
- data/lib_fqbin_src.zip +0 -0
- data/lib_fqbin_src/Makefile +66 -0
- data/lib_fqbin_src/fq +0 -0
- data/lib_fqbin_src/fq.c +165 -0
- data/lib_fqbin_src/hash_fqbin +0 -0
- data/lib_fqbin_src/hash_fqbin.c +212 -0
- data/lib_fqbin_src/idx_fqbin +21 -0
- data/lib_fqbin_src/iterate_fqbin +0 -0
- data/lib_fqbin_src/iterate_fqbin.c +136 -0
- data/lib_fqbin_src/lib_fqbin.c +1748 -0
- data/lib_fqbin_src/lib_fqbin.h +194 -0
- data/lib_fqbin_src/mk_fqbin +0 -0
- data/lib_fqbin_src/mk_fqbin.c +138 -0
- data/lib_fqbin_src/other/bwxform.c +915 -0
- data/lib_fqbin_src/other/bwxform.h +74 -0
- data/lib_fqbin_src/other/find_in_index.c +130 -0
- data/lib_fqbin_src/other/hash_fbin_nogzchunks.c +164 -0
- data/lib_fqbin_src/other/idx_fqbin +0 -0
- data/lib_fqbin_src/other/idx_fqbin.c +67 -0
- data/lib_fqbin_src/other/make_hsh.sh +14 -0
- data/lib_fqbin_src/other/rd_extras_fbin.c +45 -0
- data/lib_fqbin_src/read_fq +0 -0
- data/lib_fqbin_src/read_fq.c +143 -0
- data/lib_fqbin_src/read_fqbin +0 -0
- data/lib_fqbin_src/read_fqbin.c +101 -0
- data/lib_fqbin_src/sort_index +9 -0
- data/lib_fqbin_src/test.rb +13 -0
- data/scbi_fqbin.gemspec +25 -0
- data/test/build.rake +15 -0
- data/test/fbinfile +0 -0
- data/test/fbinfile.index +0 -0
- data/test/no_test_fill_file.rb +66 -0
- data/test/old/app.rb +43 -0
- data/test/old/bin/iterate_fastabin.rb +54 -0
- data/test/old/bin/mk_fastabin.rb +22 -0
- data/test/old/bin/rd_fastabin.rb +36 -0
- data/test/old/bin/rd_fq.rb +20 -0
- data/test/old/bioruby.rb +27 -0
- data/test/old/c/Makefile +34 -0
- data/test/old/c/fbin_lib.zip +0 -0
- data/test/old/c/iterate_fbin.c +54 -0
- data/test/old/c/libreria_gz.c +707 -0
- data/test/old/c/libreria_gz.h +127 -0
- data/test/old/c/main.c +86 -0
- data/test/old/c/mk_fbin.c +24 -0
- data/test/old/c/rd_seq_fbin.c +44 -0
- data/test/old/c/test_ffi/a.out +0 -0
- data/test/old/c/test_ffi/app.c +26 -0
- data/test/old/c/test_ffi/app.rb +19 -0
- data/test/old/c/test_ffi/liblibreria_gz.dylib +0 -0
- data/test/old/c/test_ffi/libmylibrary.dylib +0 -0
- data/test/old/c/test_ffi/my_library.rb +23 -0
- data/test/old/c/test_ffi/mylibrary.c +22 -0
- data/test/old/c/test_ffi/mylibrary.h +6 -0
- data/test/old/c/usage_instructions.txt +62 -0
- data/test/old/ext/Makefile +187 -0
- data/test/old/ext/Makefile.dario +34 -0
- data/test/old/ext/extconf.rb +8 -0
- data/test/old/ext/mk_fbin.c +24 -0
- data/test/old/ext/sample/extras.txt +4 -0
- data/{.gemtest → test/old/ext/sample/extras2.txt} +0 -0
- data/test/old/ext/sample/f1.fasta +10 -0
- data/test/old/ext/sample/f1.fasta.qual +10 -0
- data/test/old/ext/sample/f1.fbin +0 -0
- data/test/old/ext/sample/f1.fbin.index +0 -0
- data/test/old/ext/sample/main.c +86 -0
- data/test/old/ext/usage_instructions.txt +62 -0
- data/test/old/t_scbi_fastabin.rb +140 -0
- data/test/read_tests/10-original_sizes.sh +16 -0
- data/test/read_tests/20-fq_time.sh +23 -0
- data/test/read_tests/30-fbin_read_time.sh +23 -0
- data/test/read_tests/40-bsc_read_time.sh +21 -0
- data/test/read_tests/50-fq_time_x4.sh +25 -0
- data/test/read_tests/60-fbin_read_time_x4.sh +24 -0
- data/test/read_tests/70-bsc_read_time_x4.sh +32 -0
- data/test/results_bio_scbi_fasta.txt +11 -0
- data/test/{test_scbi_fbin_file.rb → scbi_fbin_file_test.rb} +0 -0
- data/test/speed.txt +81 -0
- data/test/t_scbi_fasta.rb +12 -0
- data/test/write_tests/10-original_sizes.sh +16 -0
- data/test/write_tests/20-zip_time.sh +17 -0
- data/test/write_tests/30-mk_fbin_time.sh +23 -0
- data/test/write_tests/31-mk_fbin_time_f30.sh +21 -0
- data/test/write_tests/40-gzip_time.sh +16 -0
- data/test/write_tests/41-bsc_time.sh +16 -0
- data/test/write_tests/50-zip_sizes.sh +16 -0
- data/test/write_tests/60-fbin_sizes.sh +17 -0
- data/test/write_tests/61-fbin_sizes_f30.sh +16 -0
- data/test/write_tests/70-gzip_sizes.sh +17 -0
- data/test/write_tests/80-bsc_sizes.sh +17 -0
- data/website/index.html +87 -0
- data/website/index.txt +81 -0
- data/website/javascripts/rounded_corners_lite.inc.js +285 -0
- data/website/stylesheets/screen.css +159 -0
- data/website/template.html.erb +50 -0
- metadata +208 -95
- data/History.txt +0 -19
- data/Manifest.txt +0 -12
- data/PostInstall.txt +0 -7
- data/script/console +0 -10
- data/script/destroy +0 -14
- data/script/generate +0 -14
@@ -0,0 +1,127 @@
|
|
1
|
+
#include <zlib.h>
|
2
|
+
|
3
|
+
#define VERSION 1
|
4
|
+
#define SUBVERSION 0
|
5
|
+
|
6
|
+
struct file_data {
|
7
|
+
char name[10000];
|
8
|
+
char index_name[10000];
|
9
|
+
gzFile gzf_bin;
|
10
|
+
// int file_bin;
|
11
|
+
gzFile gzf_index;
|
12
|
+
// int file_index;
|
13
|
+
// char file_outname[10000];
|
14
|
+
long long pos_chunk_gz;
|
15
|
+
// Contains the version and subversion of this file
|
16
|
+
int version;
|
17
|
+
int subversion;
|
18
|
+
// bin_search is true when a binary search can be used.
|
19
|
+
int bin_search;
|
20
|
+
// Counts the number of sequences written to the bin file, so it can
|
21
|
+
// decide where to create a new gz chunk
|
22
|
+
long long counter;
|
23
|
+
// If there is an error it is stored here so it can be retrieved.
|
24
|
+
int error;
|
25
|
+
};
|
26
|
+
|
27
|
+
// two modes:
|
28
|
+
// 1 .- new files
|
29
|
+
// 2 .- add data to files, if they don't exist they are created
|
30
|
+
int initialize_writes(struct file_data ** file, char *output_name, int mode);
|
31
|
+
|
32
|
+
/* write_seq writes a sequence to the files f_bin and its index to f_index
|
33
|
+
pos_chunk_gz is the offset of the beggining of the current gz chunk inside the file
|
34
|
+
seq_name is a pointer to the name of the sequence
|
35
|
+
fasta, quanta and extras are pointers to strings, must be zero terminated.
|
36
|
+
Returns 0 if all goes fine.
|
37
|
+
*/
|
38
|
+
|
39
|
+
void inspect_file_data_struct(struct file_data *file);
|
40
|
+
|
41
|
+
// int write_seq(gzFile *f_bin, FILE *f_index, long pos_chunk_gz, char *seq_name, char *fasta, char *quanta, char *extras);
|
42
|
+
int write_seq(struct file_data *file, char *seq_name, char *fasta, char *qual, char *extras);
|
43
|
+
|
44
|
+
int close_writes(struct file_data *file);
|
45
|
+
|
46
|
+
/*
|
47
|
+
read_seq reads from filename the sequence named seq_name and returns its
|
48
|
+
fasta, quanta and extras in those variables.
|
49
|
+
It returns 0 if there are no errors, otherwise it returns:
|
50
|
+
-2 : error opening index file (it doesn't exists)
|
51
|
+
-3 : error reading index file
|
52
|
+
-4 : error sequence not found in index file
|
53
|
+
-5 : error opening file (it doesn't exists)
|
54
|
+
-6 : error reading file
|
55
|
+
-7 : error sequence not found
|
56
|
+
-8 : error uncompressing sequence
|
57
|
+
-9 : EOF
|
58
|
+
*/
|
59
|
+
int read_seq(char *filename, char *seq_name, char **fasta, char **quanta, char **extras);
|
60
|
+
|
61
|
+
// For doing sequential reads of the whole file:
|
62
|
+
// int initialize_sequential_reads(struct file_data *filed, char *filename);
|
63
|
+
int initialize_sequential_reads(struct file_data ** filed, char *filename);
|
64
|
+
|
65
|
+
// return -9 on EOF
|
66
|
+
int read_data_sequential(struct file_data *filed, char **seq_name, char **fasta, char **qual, char **extras);
|
67
|
+
int close_sequential_reads(struct file_data *filed);
|
68
|
+
|
69
|
+
|
70
|
+
/* process_biofile reads from fname (and fname.quanta) and writes to outname
|
71
|
+
(and outname.index) with the binary format
|
72
|
+
Returns 0 if all goes fine */
|
73
|
+
int process_biofile(char *fname,char *qfname, char *efname, char *outname);
|
74
|
+
|
75
|
+
/*
|
76
|
+
Format definition
|
77
|
+
|
78
|
+
Main file that contains chunks compressed in gz
|
79
|
+
For each sequence the information of that sequence is written with the format:
|
80
|
+
28F143CJN01EBIJN 105 312 0
|
81
|
+
|
82
|
+
That is:
|
83
|
+
4 chars for the size of this header, excluding itself, that is, it is the size of
|
84
|
+
the rest of the header
|
85
|
+
sequence name
|
86
|
+
fasta size
|
87
|
+
qual size
|
88
|
+
extras size
|
89
|
+
|
90
|
+
The First sequence can be a special sequence with metainfo for this file:
|
91
|
+
30UMACOMPRESSEDFORMAT_version 0 0 0
|
92
|
+
27UMACOMPRESSEDFORMAT_1 0 0 0
|
93
|
+
|
94
|
+
|
95
|
+
|
96
|
+
Index file
|
97
|
+
|
98
|
+
Compressed using chunks
|
99
|
+
|
100
|
+
At the beggining a special sequence can be used to store metadata
|
101
|
+
like the number of fields, if a binary search can be used, etc.
|
102
|
+
|
103
|
+
That sequence will be:
|
104
|
+
UMACOMPRESSEDFORMAT version binary_search begin_of_sequential_index 0 0
|
105
|
+
|
106
|
+
If binary_search is yes then a metaindex follows to do a fast access to the
|
107
|
+
index data.
|
108
|
+
That will be the first sequence of each chunk and its offset inside the file.
|
109
|
+
(Or perphaps it can be put in another file....)
|
110
|
+
|
111
|
+
|
112
|
+
The rest of the index file will be indexes to the stored sequences, with
|
113
|
+
the following fields separated by spaces:
|
114
|
+
|
115
|
+
F143CJN01ETK00 0 471
|
116
|
+
|
117
|
+
Sequence name
|
118
|
+
begin of the compressed chunk
|
119
|
+
offset inside the chunk of the header of that sequence.
|
120
|
+
|
121
|
+
*/
|
122
|
+
|
123
|
+
|
124
|
+
|
125
|
+
|
126
|
+
|
127
|
+
|
data/test/old/c/main.c
ADDED
@@ -0,0 +1,86 @@
|
|
1
|
+
#include "libreria_gz.h"
|
2
|
+
#include <stdio.h>
|
3
|
+
#include <ctype.h>
|
4
|
+
|
5
|
+
|
6
|
+
/*******************************************************/
|
7
|
+
/* main */
|
8
|
+
/*******************************************************/
|
9
|
+
int main(int argc, char *argv[])
|
10
|
+
{
|
11
|
+
|
12
|
+
char *fasta=NULL;
|
13
|
+
char *qual=NULL;
|
14
|
+
char *extras=NULL;
|
15
|
+
int size=5000;
|
16
|
+
|
17
|
+
if (argc!=4)
|
18
|
+
{
|
19
|
+
printf("Usage %s fasta_file qual_file output_file\n\n",argv[0]);
|
20
|
+
exit(-1);
|
21
|
+
}
|
22
|
+
|
23
|
+
|
24
|
+
|
25
|
+
// prueba de lectura
|
26
|
+
printf ("Comienzo\n");
|
27
|
+
|
28
|
+
//int res=read_seq("data/F143CJN01.fbin","F143CJN01D96I9", &fasta, &qual, &extras);
|
29
|
+
//int res=read_seq("fasta_filt_w_dict.fbin","F143CJN01D96I9", &fasta, &qual, &extras);
|
30
|
+
|
31
|
+
// prueba de escritura
|
32
|
+
/*int res=write_seq("borrar","S143CJN01D96I9", "primero","QUANTAQUANTAQUANTAQUANTAQUANTA","");
|
33
|
+
res=write_seq("borrar","3CJN01D96I9", "el de enmedio","QUANTAQUANTAQUANTAQUANTAQUANTA","");
|
34
|
+
res=write_seq("borrar","3CJsdasdaN01D96I9", "ultimo","QUANTAQUANTAQUANTAQUANTAQUANTA","");
|
35
|
+
*/
|
36
|
+
//init_dicts(dict_fasta,dict_qual,32767);
|
37
|
+
//int res=process_biofile("data/F143CJN01.fasta","/tmp/prueba2gz.fbin");
|
38
|
+
int res=process_biofile(argv[1],argv[3]);
|
39
|
+
|
40
|
+
//res=read_seq("fasta_filt_w_dict.fbin","F143CJN01D96I9", &fasta, &qual, &extras);
|
41
|
+
/*
|
42
|
+
int res=read_seq("prueba2.fbin","F143CJN01DI5MZ", &fasta, &qual, &extras);
|
43
|
+
printf ("-------------------------------------------------------------\n");
|
44
|
+
printf ("RES of read_seq1 call is :%d\n",res);
|
45
|
+
if ( res==0 ) printf ("fasta:%s\n size:%d\n",fasta,sizeof(fasta));
|
46
|
+
if ( res==0 ) printf ("qual:%s\n",qual);
|
47
|
+
if (( res==0 )&& (extras!=NULL)) printf ("extras:%s\n",extras);
|
48
|
+
if ( fasta!=NULL ) {free(fasta);fasta=NULL;}
|
49
|
+
if ( qual!=NULL ) {free(qual);qual=NULL;}
|
50
|
+
if ( extras!=NULL ) {free(extras);extras=NULL;}
|
51
|
+
*/
|
52
|
+
//int res=read_seq("/tmp/prueba2gz.fbin","F143CJN01BO14N", &fasta, &qual, &extras);
|
53
|
+
//int res=read_seq("/tmp/prueba2gz.fbin","F143CJN01D2X26", &fasta, &qual, &extras);
|
54
|
+
//res=read_seq("prueba2gz.fbin","F143CJN01EBIJN", &fasta, &qual, &extras);
|
55
|
+
res=read_seq(argv[3],"F143CJN01DZW7L", &fasta, &qual, &extras);
|
56
|
+
//res=read_seq("prueba2gz.fbin","F143CJN01EN6AH", &fasta, &qual, &extras);
|
57
|
+
printf ("-------------------------------------------------------------\n");
|
58
|
+
printf ("RES of read_seq2 call is :%d\n",res);
|
59
|
+
if ( res==0 ) printf ("fasta:%s\n fasta size:%d\n",fasta,strlen(fasta));
|
60
|
+
if ( res==0 ) printf ("qual:%s\n",qual);
|
61
|
+
if (( res==0 )&& (extras!=NULL)) printf ("extras:%s\n",extras);
|
62
|
+
if ( fasta!=NULL ) {free(fasta);fasta=NULL;}
|
63
|
+
if ( qual!=NULL ) {free(qual);qual=NULL;}
|
64
|
+
if ( extras!=NULL ) {free(extras);extras=NULL;}
|
65
|
+
|
66
|
+
printf ("***************************\n");
|
67
|
+
printf ("Sequential reads\n");
|
68
|
+
|
69
|
+
initialize_sequential_reads(argv[3]);
|
70
|
+
char *sname=NULL;
|
71
|
+
while (read_data_sequential(&sname, &fasta, &qual, &extras)==0)
|
72
|
+
{
|
73
|
+
printf ("***************************\n");
|
74
|
+
printf ("RES of read_seq2 call is :%d, sname:%s\n",res,sname);
|
75
|
+
if ( res==0 ) printf ("fasta:%s fasta size:%d\n",fasta,strlen(fasta));
|
76
|
+
if ( res==0 ) printf ("qual:%s",qual);
|
77
|
+
if (( res==0 )&& (extras!=NULL)) printf ("extras:%s\n",extras);
|
78
|
+
if ( fasta!=NULL ) {free(fasta);fasta=NULL;}
|
79
|
+
if ( qual!=NULL ) {free(qual);qual=NULL;}
|
80
|
+
if ( extras!=NULL ) {free(extras);extras=NULL;}
|
81
|
+
}
|
82
|
+
close_sequential_reads();
|
83
|
+
|
84
|
+
return res;
|
85
|
+
}
|
86
|
+
|
@@ -0,0 +1,24 @@
|
|
1
|
+
#include "libreria_gz.h"
|
2
|
+
#include <stdio.h>
|
3
|
+
#include <ctype.h>
|
4
|
+
|
5
|
+
|
6
|
+
/*******************************************************/
|
7
|
+
/* main */
|
8
|
+
/*******************************************************/
|
9
|
+
int main(int argc, char *argv[])
|
10
|
+
{
|
11
|
+
// check params
|
12
|
+
if (argc!=5)
|
13
|
+
{
|
14
|
+
printf("Usage %s fasta_file qual_file extras_file output_file\n\n",argv[0]);
|
15
|
+
return -1;
|
16
|
+
}
|
17
|
+
|
18
|
+
// process file
|
19
|
+
// int res=process_biofile(argv[1],argv[2],argv[3],argv[4]);
|
20
|
+
int res=process_biofile(argv[1],argv[2],argv[3] ,argv[4]);
|
21
|
+
|
22
|
+
return res;
|
23
|
+
}
|
24
|
+
|
@@ -0,0 +1,44 @@
|
|
1
|
+
#include "libreria_gz.h"
|
2
|
+
#include <stdio.h>
|
3
|
+
#include <ctype.h>
|
4
|
+
#include <string.h>
|
5
|
+
#include <stdlib.h>
|
6
|
+
|
7
|
+
|
8
|
+
/*******************************************************/
|
9
|
+
/* main */
|
10
|
+
/*******************************************************/
|
11
|
+
int main(int argc, char *argv[])
|
12
|
+
{
|
13
|
+
|
14
|
+
char *fasta=NULL;
|
15
|
+
char *qual=NULL;
|
16
|
+
char *extras=NULL;
|
17
|
+
int size=5000;
|
18
|
+
int res=0;
|
19
|
+
|
20
|
+
if (argc!=3)
|
21
|
+
{
|
22
|
+
printf("Usage %s fbin_file seq_name\n\n",argv[0]);
|
23
|
+
return -1;
|
24
|
+
}
|
25
|
+
|
26
|
+
|
27
|
+
|
28
|
+
res=read_seq(argv[1],argv[2], &fasta, &qual, &extras);
|
29
|
+
|
30
|
+
if (res==0){
|
31
|
+
|
32
|
+
printf(">%s\n%s\n", argv[2], fasta);
|
33
|
+
printf("%s\n",qual);
|
34
|
+
if (extras!=NULL) printf ("extras:%s\n",extras);
|
35
|
+
}
|
36
|
+
|
37
|
+
//res=read_seq("prueba2gz.fbin","F143CJN01EN6AH", &fasta, &qual, &extras);
|
38
|
+
if ( fasta!=NULL ) {free(fasta);fasta=NULL;}
|
39
|
+
if ( qual!=NULL ) {free(qual);qual=NULL;}
|
40
|
+
if ( extras!=NULL ) {free(extras);extras=NULL;}
|
41
|
+
|
42
|
+
return res;
|
43
|
+
}
|
44
|
+
|
Binary file
|
@@ -0,0 +1,26 @@
|
|
1
|
+
|
2
|
+
#include <stdio.h>
|
3
|
+
#include <stdlib.h>
|
4
|
+
|
5
|
+
#include "mylibrary.h"
|
6
|
+
|
7
|
+
int main()
|
8
|
+
{
|
9
|
+
double c, d ;
|
10
|
+
int errcode ;
|
11
|
+
struct SomeObject *objptr ;
|
12
|
+
|
13
|
+
c = calculate_something(42, 98.6);
|
14
|
+
|
15
|
+
if ((errcode = error_code()) != 0) {
|
16
|
+
fprintf(stderr, "error calculating something: %d\n", errcode);
|
17
|
+
exit(1);
|
18
|
+
}
|
19
|
+
|
20
|
+
objptr = create_object("my object") ;
|
21
|
+
d = calculate_something_else(c, objptr) ;
|
22
|
+
free_object(objptr) ;
|
23
|
+
|
24
|
+
fprintf(stdout, "calculated %f\n", d);
|
25
|
+
exit(0) ;
|
26
|
+
}
|
@@ -0,0 +1,19 @@
|
|
1
|
+
require 'my_library'
|
2
|
+
|
3
|
+
o = MyLibrary.new
|
4
|
+
|
5
|
+
r=o.calculate_something(42,98.6)
|
6
|
+
puts r
|
7
|
+
|
8
|
+
# c = MyLibrary.calculate_something(42, 98.6) # note FFI handles literals just fine
|
9
|
+
#
|
10
|
+
# if ( (errcode = MyLibrary.error_code()) != 0)
|
11
|
+
# puts "error calculating something: #{errcode}"
|
12
|
+
# exit 1
|
13
|
+
# end
|
14
|
+
#
|
15
|
+
# objptr = MyLibrary.create_object("my object") # note FFI handles string literals as well
|
16
|
+
# d = MyLibrary.calculate_something_else(c, objptr)
|
17
|
+
# MyLibrary.free_object(objptr)
|
18
|
+
#
|
19
|
+
# puts "calculated #{d}"
|
Binary file
|
Binary file
|
@@ -0,0 +1,23 @@
|
|
1
|
+
require 'ffi'
|
2
|
+
|
3
|
+
module MyModule
|
4
|
+
extend FFI::Library
|
5
|
+
ffi_lib "mylibrary"
|
6
|
+
|
7
|
+
attach_function :calculate_something, [:int, :float], :double
|
8
|
+
attach_function :error_code, [], :int # note empty array for functions taking zero arguments
|
9
|
+
attach_function :create_object, [:string], :pointer
|
10
|
+
attach_function :calculate_something_else, [:double, :pointer], :double
|
11
|
+
attach_function :free_object, [:pointer], :void
|
12
|
+
|
13
|
+
end
|
14
|
+
|
15
|
+
class MyLibrary
|
16
|
+
|
17
|
+
include MyModule
|
18
|
+
|
19
|
+
def initialize
|
20
|
+
puts calculate_something(1,2.3)
|
21
|
+
end
|
22
|
+
|
23
|
+
end
|
@@ -0,0 +1,22 @@
|
|
1
|
+
#include "mylibrary.h"
|
2
|
+
|
3
|
+
double calculate_something(int a, float b){
|
4
|
+
return a+b;
|
5
|
+
}
|
6
|
+
|
7
|
+
int error_code(void){
|
8
|
+
return 0;
|
9
|
+
}
|
10
|
+
|
11
|
+
struct SomeObject* create_object(char* name){
|
12
|
+
// SomeObject e;
|
13
|
+
return 0;
|
14
|
+
}
|
15
|
+
|
16
|
+
double calculate_something_else(double c, struct SomeObject* obj){
|
17
|
+
return 1.1;
|
18
|
+
}
|
19
|
+
|
20
|
+
void free_object(void* pointer_to_memory){
|
21
|
+
|
22
|
+
}
|
@@ -0,0 +1,62 @@
|
|
1
|
+
|
2
|
+
Manual of the YAGFF (Yet Another Genetic File Format)
|
3
|
+
OGF (Optimized Genetic Format)
|
4
|
+
|
5
|
+
|
6
|
+
First a call to :
|
7
|
+
error=initialize_writes(outname,1);
|
8
|
+
|
9
|
+
Is needed to initialize the files.
|
10
|
+
|
11
|
+
Its arguments are:
|
12
|
+
|
13
|
+
- the name of the main output file and to this the suffix .index will be
|
14
|
+
appended for the index file.
|
15
|
+
|
16
|
+
- The mode of the write:
|
17
|
+
1 : create a new file
|
18
|
+
2 : append data to already existing files
|
19
|
+
|
20
|
+
It returns
|
21
|
+
0 if everything goes fine,
|
22
|
+
-1 if there is a problem with already existing files, like an incorrect
|
23
|
+
version number
|
24
|
+
-2 if there is a problem creating or opening files
|
25
|
+
|
26
|
+
In case of error it writes it on stderr
|
27
|
+
|
28
|
+
|
29
|
+
Then request to write can be done:
|
30
|
+
|
31
|
+
int error_wr=write_seq(sname, fasta,qual,extras);
|
32
|
+
|
33
|
+
Where sname is a char* with the name of the sequence, and fasta, qual and
|
34
|
+
extras are char* with the fasta, qual and extras data.
|
35
|
+
|
36
|
+
At the end a call to : closes_writes();
|
37
|
+
is needed to close the opened files
|
38
|
+
|
39
|
+
|
40
|
+
|
41
|
+
The reads are even easier:
|
42
|
+
|
43
|
+
int read_seq(char *filename, char *seq_name, char **fasta, char **qual, char **extras)
|
44
|
+
The arguments are:
|
45
|
+
filename: is a char* used as input the name of the main file, .index is appended for the index file
|
46
|
+
seq_name : char* that contains the name of the searched sequence
|
47
|
+
|
48
|
+
And the result are returned in :
|
49
|
+
fasta, qual and extras, if they point to NULL they are malloced to bufsize=150000 bytes (defined inside the function)
|
50
|
+
else they are used without checking its size
|
51
|
+
|
52
|
+
|
53
|
+
|
54
|
+
For doing sequential reads an untested procedure is implemented:
|
55
|
+
|
56
|
+
int initialize_sequential_reads(char *filename);
|
57
|
+
|
58
|
+
// return -9 on EOF
|
59
|
+
int read_data_sequential(char **seq_name, char **fasta, char **qual, char **extras)
|
60
|
+
int close_sequential_reads();
|
61
|
+
|
62
|
+
|