/*********************************************************************
** METHOD :
** PURPOSE :
** INPUT :
** OUTPUT :
** RETURN :
** REMARKS : Warning! this method isn't exactly basename() as you'd expect in C++. It returns the base name but cuts everything after the last dot.
*********************************************************************/
IFileSystem::Path FileSystemCommon::getBaseName (const Path& path, bool cutToFirstDot)
{
/** We duplicate the provided path. */
char* reads_path = strdup (path.c_str());
/** We build the basename; it still may have a suffix. */
std::string reads_name (basename(reads_path)); // posix basename() may alter reads_path
/** We release the duplicated path. */
free (reads_path);
//string prefix = System::file().getBaseName(_inputFilename);;
while (reads_name.find('.') != string::npos){ // make sure there is a dot in the file, else the basename is the file itself
/** We look for the beginnin of the suffix. */
int lastindex = reads_name.find_last_of(".");
/** We build the result. */
reads_name = reads_name.substr(0, lastindex);
if (cutToFirstDot == false)
break;
}
//int lastindex = reads_name.find_last_of (".");
//Path result = reads_name.substr(0, lastindex);
/** We return the base name, without suffix. */
return reads_name;
}
int main(int argc, char *argv[])
{
if(argc < 3)
{
fprintf (stderr,"%s: [d]isk [s]treaming of [k]-mers (constant-memory k-mer counting)\n",argv[0]);
fprintf (stderr,"usage:\n");
fprintf (stderr," %s input_file kmer_size [-t min_abundance] [-m max_memory] [-d max_disk_space] [-o out_prefix] [-histo]\n",argv[0]);
fprintf (stderr,"details:\n [-t min_abundance] filters out k-mers seen ( < min_abundance ) times, default: 1 (all kmers are returned)\n [-m max_memory] is in MB, default: min(total system memory / 2, 5 GB) \n [-d max_disk_space] is in MB, default: min(available disk space / 2, reads file size)\n [-o out_prefix] saves results in [out_prefix].solid_kmers. default out_prefix = basename(input_file)\n [-histo] outputs histogram of kmers abundance\n [-rev] outputs only one of forward or reverse complement k-mers\n Input file can be fasta, fastq, gzipped or not, or a file containing a list of file names.\n");
#ifdef SVN_REV
fprintf(stderr,"Running dsk version %s\n",STR(SVN_REV));
#endif
return 0;
}
// reads file
Bank *Reads = new Bank(argv[1]);
if (argv[2][0] == '-')
{
printf("please specify a k value\n");
exit(1);
}
/* Changes by Raunaq
* Code addition for taking in multiple values of k. The file containing should have values of k sorted in decreasing order
*
*/
int *Kmerlist = loadKmers(argv[2]);
// kmer size
//fprintf(stderr,"Smallest kmer is %d \n",smallestKmer);
sizeKmer = Kmerlist[0];
if (sizeKmer>(int)(sizeof(kmer_type)*4))
{
printf("Max kmer size on this compiled version is %lu\n",sizeof(kmer_type)*4);
exit(1);
}
kmerMask=(((kmer_type)1)<<(sizeKmer*2))-1;
// default solidity
nks = 1;
// default max memory
max_memory = 5*1024;
#ifndef OSX
struct sysinfo info;
sysinfo(&info);
int total_ram = (int)(((double)info.totalram*(double)info.mem_unit)/1024/1024);
printf("Total RAM: %d MB\n",total_ram);
#else
int total_ram = 128*1024;
#endif
// default prefix is the reads file basename
char *reads_path=strdup(argv[1]);
string reads_name(basename(reads_path)); // posix basename() may alter reads_path
free(reads_path);
int lastindex = reads_name.find_last_of(".");
strcpy(prefix,reads_name.substr(0, lastindex).c_str());
for (int n_a = 3; n_a < argc ; n_a++)
{
if (strcmp(argv[n_a],"-t")==0)
nks = atoi(argv[n_a+1]);
if (strcmp(argv[n_a],"-o")==0)
strcpy(prefix,argv[n_a+1]);
}
int verbose = 0;
bool reverse = false;
max_disk_space = 0;
output_histo =false;
// parse the remaining arguments: these will override the default max memory / max disk
for (int n_a = 3; n_a < argc ; n_a++)
{
if (strcmp(argv[n_a],"-m")==0)
max_memory = atoi(argv[n_a+1]);
if (strcmp(argv[n_a],"-d")==0)
max_disk_space = atoi(argv[n_a+1]);
if (strcmp(argv[n_a],"-v")==0)
verbose = 1;
if (strcmp(argv[n_a],"-vv")==0)
verbose = 2;
if (strcmp(argv[n_a],"-histo")==0)
output_histo =true;
if (strcmp(argv[n_a],"-rev")==0)
reverse = true;
}
if (max_memory > total_ram)
{
printf("Maximum memory (%d MB), exceeds total RAM (%d MB). Setting maximum memory to %d MB.\n",max_memory,total_ram,total_ram/2);
max_memory = total_ram/2;
}
STARTWALL(0);
sorting_count(Reads,prefix,max_memory,max_disk_space,true,verbose,reverse);
STOPWALL(0,"Total");
delete Reads;
return 0;
}
