string usage(ArgProcessor args)
{
bool show_advanced = args.isArgSet("--show_advanced");
stringstream usage_info;
usage_info
<< endl << endl
<< "Usage:" << endl
<< " inchworm --reads <filename.fasta> --run_inchworm [opts] " << endl
<< " inchworm --kmers <filename.fasta> --run_inchworm [opts] " << endl
<< " inchworm --kmer_files_listing <kmer_file_list.txt> --run_inchworm [opts] " << endl << endl
<< "**Required" << endl
<< " --reads <str> " << ":fasta file containing reads" << endl
<< " --kmers <str> " << ":fasta file containing kmers" << endl
<< " --kmer_files_listing <str> " << ":file listing filenames containing kmers" << endl
<< endl;
usage_info
<< endl
<< "** Run modes:" << endl
<< " --run_inchworm " << ":run inchworm, report sequences" << endl;
if (show_advanced) {
usage_info
<< " * advanced: " << endl
<< " --describe_kmers " << ":just describe kmers in the graph" << endl
<< " --checkFastaPath <str> " << ":check fastea seqs for path in kmer graph" << endl
<< " --threadFasta <str> " << ":thread sequences through the kmer graph, report kmer" << endl;
}
usage_info
<< endl
<< "** General opts" << endl
<< " -K <int> "
<< ":kmer length (default: 25, meaning 24mer overlaps) max = 32 (stored as 64-bit integers, 2-base char encoding)"
<< endl
<< " --minKmerCount <int> "
<< ":min kmer count, lower abundant kmers are pruned from graph (default: 1)" << endl
<< " -L <int> " << ":min contig length to be reported (default: 25)" << endl
<< " --min_assembly_coverage <int> "
<< ":min kmer coverage of an assembled contig to be reported (default: 2)" << endl
<< " --coverage_outfile <str> " << ":file to store per-base kmer coverage for contigs" << endl
<< " --DS " << ":double-stranded RNA-Seq mode (not strand-specific)" << endl
<< " --no_prune_error_kmers "
<< ":disable pruning of kmers that occur at below --min_ratio_non_error " << endl
<< " --min_ratio_non_error <float> "
<< ":min ratio for low/high alternative extension that is not an error (default: 0.05)" << endl
<< " --num_threads <int> "
<< ":number of threads to use. (by default, uses OMP_NUM_THREADS env var setting)" << endl
<< " --PARALLEL_IWORM "
<< ":run the contig building in parallel; by default, only does parallel read parsing." << endl;
if (show_advanced) {
usage_info
<< " * advanced opts:" << endl
// << " -R <int> " << ":maximum recursion length (default: 1)" << endl
<< " --min_any_entropy <float> "
<< ":kmer must have at least this entropy to be added to the graph (default: 0)" << endl
<< " --min_seed_entropy <float> " << ":min seed kmer entropy (default: 1.5)" << endl
<< " --min_seed_coverage <int> " << ":min seed kmer coverage (default: 2)" << endl
// << " --crawl " << ":drag tail of inchworm by --crawl_length" << endl
// << " --crawl_length <int> " << ":length of tail end crawl (default: 1)" << endl
// << " --pacman " << ":exclude kmers as they're visited in descending order of counts" << endl
// << " --pruneSingletonKmersReadInterval <long> " << ":at num reads parsed interval, prune singleton kmers (default: 0)" << endl
// << " --min_con_ratio <float> " << ":minimum connectivity ratio (min/max) between neighboring kmers (default: 0)" << endl
// << " --reassembleIworm " << ":redo assembly using previous inchworm outputs. Post-process to vary-K" << endl;
<< " * developer flags: " << endl
<< " --DEVEL_no_kmer_sort "
<< ":no sorting of kmer by abundance values occurs. Kmer seeds chosen according to hash iterator order."
<< endl
<< " --DEVEL_no_greedy_extend "
<< ":instead of greedy contig extensions, any non-zero kmer is chosen to extend." << endl
<< " --DEVEL_no_tie_breaking " << ":if tie, choose either path randomly" << endl
<< " --DEVEL_zero_kmer_on_use " << ":zero kmer counts on use rather than after contig construction"
<< endl
<< " --DEVEL_rand_fracture_prob <float> "
<< " :probability of fracturing an inchworm contig at each extension (default: 0 (off)" << endl
<< " --SINGLE_PHASE "
<< ":in parallel mode, do a single iworm construction phase instead of default 2-phase" << endl
<< " --keep_tmp_files " << ":retain temporary files." << endl;
}
usage_info
<< endl
<< "** Misc opts." << endl
<< " --help " << ":produce this menu." << endl
<< " --monitor <int> " << ":verbosity. ( '1': recommended, '2': for debugging ) " << endl
<< " --show_advanced " << ":more advanced options (mostly experimental)" << endl
<< endl << endl << endl;
return (usage_info.str());
}
string usage (ArgProcessor args) {
bool show_advanced = args.isArgSet("--show_advanced");
stringstream usage_info;
usage_info
<< endl << endl
<< "Usage:" << endl
<< " inchworm --reads <filename.fasta> --run_inchworm [opts] " << endl
<< " inchworm --kmers <filename.fasta> --run_inchworm [opts] " << endl
<< " inchworm --kmer_files_listing <kmer_file_list.txt> --run_inchworm [opts] " << endl << endl
<< "**Required" << endl
<< " --reads <str> " << ":fasta file containing reads" << endl
<< " --kmers <str> " << ":fasta file containing kmers" << endl
<< " --kmer_files_listing <str> " << ":file listing filenames containing kmers" << endl
<< endl;
usage_info
<< endl
<< "** Run modes:" << endl
<< " --run_inchworm " << ":run inchworm, report sequences" << endl;
if (show_advanced) {
usage_info
<< " * advanced: " << endl
// << " --describe_kmers " << ":just describe kmers in the graph" << endl
<< " --checkFastaPath <str> " << ":check fastea seqs for path in kmer graph" << endl
<< " --threadFasta <str> " << ":thread sequences through the kmer graph, report kmer" << endl;
}
usage_info
<< endl
<< "** General opts" << endl
<< " -K <int> " << ":kmer length (default: 25, meaning 24mer overlaps) max = 32 (stored as 64-bit integers, 2-base char encoding)" << endl
<< " --minKmerCount <int> " << ":min kmer count, lower abundant kmers are pruned from graph (default: 1)" << endl
<< " -L <int> " << ":min contig length to be reported (default: 25)" << endl
<< " --min_assembly_coverage <int> " << ":min kmer coverage of an assembled contig to be reported (default: 2)" << endl
<< " --coverage_outfile <str> " << ":file to store per-base kmer coverage for contigs" << endl
<< " --DS " << ":double-stranded RNA-Seq mode (not strand-specific)" << endl
<< " --no_prune_error_kmers " << ":disable pruning of kmers that occur at below --min_ratio_non_error " << endl
<< " --min_ratio_non_error <float> " << ":min ratio for low/high alternative extension that is not an error (default: 0.05)" << endl
<< " --num_threads <int> " << ":number of threads to use." << endl
;
if (show_advanced) {
usage_info
<< " * advanced opts:" << endl
// << " -R <int> " << ":maximum recursion length (default: 1)" << endl
<< " --min_any_entropy <float> " << ":kmer must have at least this entropy to be added to the graph (default: 0)" << endl
<< " --min_seed_entropy <float> " << ":min seed kmer entropy (default: 1.5)" << endl
<< " --min_seed_coverage <int> " << ":min seed kmer coverage (default: 2)" << endl
// << " --crawl " << ":drag tail of inchworm by --crawl_length" << endl
// << " --crawl_length <int> " << ":length of tail end crawl (default: 1)" << endl
// << " --pacman " << ":exclude kmers as they're visited in descending order of counts" << endl
// << " --pruneSingletonKmersReadInterval <long> " << ":at num reads parsed interval, prune singleton kmers (default: 0)" << endl
// << " --min_con_ratio <float> " << ":minimum connectivity ratio (min/max) between neighboring kmers (default: 0)" << endl
// << " --reassembleIworm " << ":redo assembly using previous inchworm outputs. Post-process to vary-K" << endl;
;
}
usage_info
<< endl
<< "** Misc opts." << endl
<< " --help " << ":produce this menu." << endl
<< " --monitor <int> " << ":verbosity. ( '1': recommended, '2': for debugging ) " << endl
<< " --show_advanced " << ":more advanced options (mostly experimental)" << endl
<< endl << endl << endl;
return(usage_info.str());
}
