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;
}
int main(int argc, char *argv[])
{
if(argc < 6)
{
fprintf (stderr,"usage:\n");
fprintf (stderr," %s input_file kmer_size min_abundance estimated_genome_size prefix\n",argv[0]);
fprintf (stderr,"hints:\n min_abundance ~ 3\n estimated_genome_size is in bp, does not need to be accurate, only controls memory usage\n prefix is any name you want the results to start with\n");
return 1;
}
bool FOUR_BLOOM_VERSION = true;
// shortcuts to go directly to assembly using serialized bloom and serialized hash
int START_FROM_SOLID_KMERS=0; // if = 0, construct the fasta file of solid kmers, if = 1, start directly from that file
int LOAD_FALSE_POSITIVE_KMERS=0; // if = 0, construct the fasta file of false positive kmers (debloom), if = 1, load that file into the hashtable
int NO_FALSE_POSITIVES_AT_ALL=0; // if = 0, normal behavior, if = 1, don't load false positives (will be a probabilistic de bruijn graph)
int max_disk_space = 0;// let dsk decide
for (int n_a = 6; n_a < argc ; n_a++)
{
if (strcmp(argv[n_a],"--original") == 0)
FOUR_BLOOM_VERSION = false;
if (strcmp(argv[n_a],"--dont-count")==0)
START_FROM_SOLID_KMERS = 1;
if (strcmp(argv[n_a],"--dont-debloom")==0)
LOAD_FALSE_POSITIVE_KMERS = 1;
if (strcmp(argv[n_a],"--just-assemble")==0)
{
START_FROM_SOLID_KMERS = 1;
LOAD_FALSE_POSITIVE_KMERS = 1;
}
if (strcmp(argv[n_a],"--titus-mode")==0)
NO_FALSE_POSITIVES_AT_ALL = 1;
if (strcmp(argv[n_a],"-d")==0)
max_disk_space = atoi(argv[n_a+1]);
if (strcmp(argv[n_a],"-maxc")==0)
max_couv = atoi(argv[n_a+1]);
if (strcmp(argv[n_a],"--le-changement")==0)
{printf("c'est maintenant!\n");exit(0);}
}
// kmer size
sizeKmer=27; // let's make it even for now, because i havnt thought of how to handle palindromes (dont want to stop on them)
if(argc >= 3)
{
sizeKmer = atoi(argv[2]);
if (sizeKmer%2==0)
{
sizeKmer-=1;
printf("Need odd kmer size to avoid palindromes. I've set kmer size to %d.\n",sizeKmer);
}
if (sizeKmer>((int)sizeof(kmer_type)*4))
{
printf("Max kmer size on this compiled version is %lu\n",sizeof(kmer_type)*4);
exit(1);
}
}
if (sizeKmer == (int)(sizeof(kmer_type)*4))
kmerMask = -1;
else
kmerMask=(((kmer_type)1)<<(sizeKmer*2))-1;
double lg2 = log(2);
if (sizeKmer > 128)
{
FOUR_BLOOM_VERSION = false;
printf("Reverted to single Bloom filter implementation for k>128\n");
}
if (!FOUR_BLOOM_VERSION)
NBITS_PER_KMER = log(16*sizeKmer*(lg2*lg2))/(lg2*lg2); // needed to process argv[5]
else
NBITS_PER_KMER = rvalues[sizeKmer][1];
// solidity
nks =NNKS;
if(argc >= 4)
{
nks = atoi(argv[3]);
if (nks==0) nks=1; // min abundance can't be 0
}
if(argc >= 5)
{
genome_size = atoll(argv[4]);
// int estimated_bloom_size = max( (int)ceilf(log2f(genome_size * NBITS_PER_KMER )), 1);
uint64_t estimated_bloom_size = (uint64_t) (genome_size * NBITS_PER_KMER);
uint64_t estimated_nb_FP = (uint64_t)(genome_size * 4 * powf(0.6,11)); // just indicative
//max_memory = max( (1LL << estimated_bloom_size)/8LL /1024LL/1024LL, 1LL );
max_memory = max((int64_t) estimated_bloom_size/8LL /1024LL/1024LL,1LL);
printf("estimated values: nbits Bloom %lli, nb FP %lld, max memory %i MB\n",estimated_bloom_size,estimated_nb_FP,max_memory);
}
// output prefix
if(argc >= 6)
{
strcpy(prefix,argv[5]);
}
fprintf (stderr,"taille cell %lu \n", sizeof(cell<kmer_type>));
STARTWALL(0);
Bank *Reads = new Bank(argv[1]);
// counter kmers, write solid kmers to disk
if (!START_FROM_SOLID_KMERS)
{
int verbose = 0;
bool write_count = false;
bool skip_binary_conversion = false;
sorting_count(Reads,prefix,max_memory,max_disk_space,write_count,verbose, skip_binary_conversion);
}
// debloom, write false positives to disk, insert them into false_positives
if (! LOAD_FALSE_POSITIVE_KMERS)
{
debloom(order, max_memory);
}
bloo1 = bloom_create_bloo1((BloomCpt *)NULL, false);
if (! NO_FALSE_POSITIVES_AT_ALL)
{
// load false positives from disk into false_positives
if (!FOUR_BLOOM_VERSION)
false_positives = load_false_positives();
else
false_positives = load_false_positives_cascading4();
}
else
{
// titus mode: no FP's
false_positives = dummy_false_positives();
}
// return 1;
assemble();
STOPWALL(0,"Total");
delete Reads;
return 0;
}
