Exemplo n.º 1
0
static GtOptionParser* gt_sequniq_option_parser_new(void *tool_arguments)
{
  GtSequniqArguments *arguments = tool_arguments;
  GtOptionParser *op;
  GtOption *seqit_option, *verbose_option, *width_option, *rev_option,
           *nofseqs_option;
  gt_assert(arguments);

  op = gt_option_parser_new("[option ...] sequence_file [...] ",
                            "Filter out repeated sequences in given "
                            "sequence files.");

  /* -seqit */
  seqit_option = gt_option_new_bool("seqit", "use sequence iterator",
                                    &arguments->seqit, false);
  gt_option_is_development_option(seqit_option);
  gt_option_parser_add_option(op, seqit_option);

  /* -nofseqs */
  nofseqs_option = gt_option_new_uword("nofseqs", "number of sequences "
      "(improves efficiency)\ndefault: unspecified",
      &arguments->nofseqs, 0);
  gt_option_is_development_option(nofseqs_option);
  gt_option_hide_default(nofseqs_option);
  gt_option_parser_add_option(op, nofseqs_option);

  /* -rev */
  rev_option = gt_option_new_bool("rev", "also filter out sequences whose "
      "reverse complement is identical to a sequence already output",
      &arguments->rev, false);
  gt_option_parser_add_option(op, rev_option);

  /* -v */
  verbose_option = gt_option_new_verbose(&arguments->verbose);
  gt_option_parser_add_option(op, verbose_option);

  /* -width */
  width_option = gt_option_new_width(&arguments->width);
  gt_option_parser_add_option(op, width_option);

  gt_output_file_info_register_options(arguments->ofi, op, &arguments->outfp);

  /* option implications */
  gt_option_imply(verbose_option, seqit_option);

  gt_option_parser_set_comment_func(op, gt_gtdata_show_help, NULL);
  gt_option_parser_set_min_args(op, 1U);
  return op;
}
Exemplo n.º 2
0
static GtOptionParser* gt_gff3validator_option_parser_new(void *tool_arguments)
{
    GFF3ValidatorArguments *arguments = tool_arguments;
    GtOptionParser *op;
    GtOption *option;
    gt_assert(arguments);

    /* init */
    op = gt_option_parser_new("[option ...] [GFF3_file ...]",
                              "Strictly validate given GFF3 files.");

    /* typecheck options */
    gt_typecheck_info_register_options(arguments->tci, op);

    /* xrfcheck options */
    gt_xrfcheck_info_register_options(arguments->xci, op);

    /* -strict */
    option = gt_option_new_bool("strict", "be very strict during GFF3 parsing "
                                "(stricter than the specification requires)",
                                &arguments->strict, false);
    gt_option_is_development_option(option);
    gt_option_parser_add_option(op, option);

    return op;
}
Exemplo n.º 3
0
void gt_typecheck_info_register_options(GtTypecheckInfo *tci,
                                        GtOptionParser *op)
{
    GtOption *typecheck_option, *built_in_option;
    gt_assert(tci && op);
    gt_assert(!tci->typecheck_option); /* can only called once */

    /* -typecheck */
    typecheck_option =
        gt_option_new_string("typecheck", "check GFF3 types against \"id\" and "
                             "\"name\" tags in given OBO file and validate parent "
                             "(part-of) relationships.\nIf no argument is given, "
                             "the sofa.obo file from the gtdata/obo_files "
                             "directory is used.\nIf an argument is given, it is "
                             "used as an OBO filename.\nIn the case that such a "
                             "file does not exist '.obo' is added to the argument "
                             "and loading the resulting filename from the "
                             "gtdata/obo_files directory is attempted.",
                             tci->typecheck, NULL);
    gt_option_argument_is_optional(typecheck_option);
    gt_option_parser_add_option(op, typecheck_option);
    tci->typecheck_option = gt_option_ref(typecheck_option);

    /* -typecheck-built-in */
    built_in_option = gt_option_new_bool("typecheck-built-in",
                                         "use built-in type checker",
                                         &tci->typecheck_built_in, false);
    gt_option_is_development_option(built_in_option);
    gt_option_parser_add_option(op, built_in_option);
    gt_option_exclude(typecheck_option, built_in_option);
}
Exemplo n.º 4
0
void gt_typecheck_info_register_options_with_default(GtTypecheckInfo *tci,
                                                     GtOptionParser *op,
                                                     const char *default_v)
{
  GtOption *typecheck_option, *built_in_option;
  gt_assert(tci && op);
  gt_assert(!tci->typecheck_option); /* can only called once */

  /* -typecheck */
  typecheck_option =
    gt_option_new_string("typecheck", "use an ontology given in an OBO file "
                         "to validate parent-child relationships.\n"
                         "If no argument is given, "
                         "the sofa.obo file from the gtdata/obo_files "
                         "directory is used.\nIf an argument is given, it is "
                         "used as an OBO filename.\nIn the case that such a "
                         "file does not exist '.obo' is added to the argument "
                         "and loading the resulting filename from the "
                         "gtdata/obo_files directory is attempted.",
                         tci->typecheck, default_v);
  gt_option_argument_is_optional(typecheck_option);
  gt_option_parser_add_option(op, typecheck_option);
  tci->typecheck_option = gt_option_ref(typecheck_option);

  /* -typecheck-built-in */
  built_in_option = gt_option_new_bool("typecheck-built-in",
                                       "use built-in type checker",
                                       &tci->typecheck_built_in, false);
  gt_option_is_development_option(built_in_option);
  gt_option_parser_add_option(op, built_in_option);
  gt_option_exclude(typecheck_option, built_in_option);
}
static GtOptionParser*
gt_compreads_decompress_option_parser_new(void *tool_arguments)
{
  GtCsrHcrDecodeArguments *arguments = tool_arguments;
  GtOptionParser *op;
  GtOption *option;
  gt_assert(arguments);

  /* init */
  op = gt_option_parser_new("[option ...] (-file file)",
                         "Decodes a file of compressed reads.");

  option = gt_option_new_bool("v", "be verbose",
                              &arguments->verbose, false);
  gt_option_parser_add_option(op, option);

  option = gt_option_new_bool("descs","enable description decoding",
                              &arguments->descs, false);
  gt_option_parser_add_option(op, option);

  option = gt_option_new_string("file", "specify base name of files containing"
                                " HCR.",
                                arguments->file, NULL);
  gt_option_is_mandatory(option);
  gt_option_parser_add_option(op, option);

  option = gt_option_new_string("name", "specify base name for decoded hcr"
                                " (suffix will be \".fastq\")",
                                arguments->name, NULL);
  gt_option_parser_add_option(op, option);

  option = gt_option_new_string("smap", "specify file containing alphabet"
                                "description (file must be an .al1 file)."
                                " If \"-smap\" is not set, dna alphabet is"
                                " used.",
                                arguments->smap, NULL);
  gt_option_parser_add_option(op, option);

  option = gt_option_new_range("range", "decode multiple consecutive reads."
                               " If range is not specified, the "
                               " entire file will be decoded.",
                               &arguments->rng, NULL);
  gt_option_parser_add_option(op, option);

  option = gt_option_new_ulong("benchmark", "decode given number random reads "
                               "and report the time to do this",
                               &arguments->bench, 0);
  gt_option_is_development_option(option);
  gt_option_parser_add_option(op, option);

  gt_option_parser_set_min_max_args(op, 0U, 0U);
  return op;
}
Exemplo n.º 6
0
static GtOptionParser* gt_cds_option_parser_new(void *tool_arguments)
{
  CDSArguments *arguments = tool_arguments;
  GtOptionParser *op;
  GtOption *option;
  gt_assert(arguments);

  op = gt_option_parser_new("[option ...] [GFF3_file]",
                            "Add CDS (coding sequence) features to exon "
                            "features given in GFF3 file.");

  /* -minorflen */
  option = gt_option_new_uint_min("minorflen", "set the minimum length an open "
                                  "reading frame (ORF) must have to be added "
                                  "as a CDS feature (measured in amino acids)",
                                  &arguments->minorflen, 64, 1);
  gt_option_parser_add_option(op, option);

  /* -startcodon */
  option = gt_option_new_bool("startcodon", "require than an ORF must begin "
                              "with a start codon", &arguments->start_codon,
                              false);
  gt_option_parser_add_option(op, option);

  /* -finalstopcodon */
  option = gt_option_new_bool("finalstopcodon", "require that the final ORF "
                              "must end with a stop codon",
                              &arguments->final_stop_codon, false);
  gt_option_parser_add_option(op, option);

  /* -genericstartcodons */
  option = gt_option_new_bool("genericstartcodons", "use generic start codons",
                              &arguments->generic_start_codons, false);
  gt_option_is_development_option(option);
  gt_option_parser_add_option(op, option);

  /* -seqfile, -matchdesc, -usedesc and -regionmapping */
  gt_seqid2file_register_options(op, arguments->s2fi);

  /* -v */
  option = gt_option_new_verbose(&arguments->verbose);
  gt_option_parser_add_option(op, option);

  /* output file options */
  gt_outputfile_register_options(op, &arguments->outfp, arguments->ofi);

  gt_option_parser_set_comment_func(op, gt_gtdata_show_help, NULL);
  gt_option_parser_set_max_args(op, 1);

  return op;
}
Exemplo n.º 7
0
static GtOptionParser *gt_tyr_search_option_parser_new(void *tool_arguments)
{
  GtOptionParser *op;
  GtOption *option, *optiontyr, *optionqueries;
  Tyr_search_options *arguments = tool_arguments;

  op = gt_option_parser_new("[options] -tyr tallymer-index -q queryfile0 "
                            "[queryfile1..] [options]",
                            "Search a set of k-mers in an index constructed "
                            "by \"gt tyr mkindex\".");
  gt_option_parser_set_mailaddress(op,"<*****@*****.**>");

  optiontyr = gt_option_new_string("tyr","specify tallymer-index",
                                   arguments->str_inputindex,
                                   NULL);
  gt_option_is_mandatory(optiontyr);
  gt_option_parser_add_option(op, optiontyr);

  optionqueries = gt_option_new_filenamearray("q","specify query file names",
                                              arguments->queryfilenames);
  gt_option_is_mandatory(optionqueries);
  gt_option_parser_add_option(op, optionqueries);

  option = gt_option_new_string("strand",
                                "specify the strand to be searched: "
                                "use f (for forward strand) or "
                                "p (for reverse complemented strand) or "
                                "fp (for both); default is f",
                                arguments->strandspec,
                                "f");
  gt_option_parser_add_option(op, option);

  option = gt_option_new_stringarray("output",
                                     "specify output flags "
                                     "(qseqnum, qpos, counts, sequence)",
                                     arguments->showmodespec);
  gt_option_parser_add_option(op, option);

  option = gt_option_new_bool("test", "perform tests to verify program "
                                      "correctness", &arguments->performtest,
                                      false);
  gt_option_is_development_option(option);
  gt_option_parser_add_option(op, option);

  option = gt_option_new_verbose(&arguments->verbose);
  gt_option_parser_add_option(op, option);

  return op;
}
Exemplo n.º 8
0
static GtOptionParser
            *gt_cge_spacedseed_option_parser_new(void *tool_arguments)
{
  GtOptionParser *op;
  GtOption *option,
           *optionesaindex,
           *optionpckindex;
  Cge_spacedseed_options *arguments = tool_arguments;

  op = gt_option_parser_new("[options]",
                            "Match spaced seeds.");
  gt_option_parser_set_mailaddress(op,"<*****@*****.**>");

  optionesaindex = gt_option_new_string("esa",
                                     "Specify index (enhanced suffix array)",
                                     arguments->str_inputindex, NULL);
  gt_option_parser_add_option(op, optionesaindex);
  arguments->refoptionesaindex = gt_option_ref(optionesaindex);

  optionpckindex = gt_option_new_string("pck",
                                     "Specify index (packed index)",
                                     arguments->str_inputindex, NULL);
  gt_option_parser_add_option(op, optionpckindex);
  arguments->refoptionpckindex = gt_option_ref(optionpckindex);
  gt_option_exclude(optionesaindex,optionpckindex);
  gt_option_is_mandatory_either(optionesaindex,optionpckindex);

  option = gt_option_new_filenamearray("q",
                                    "Specify files containing the "
                                    "query sequences",
                                    arguments->queryfilenames);
  gt_option_parser_add_option(op, option);
  gt_option_is_mandatory(option);

  option = gt_option_new_bool("cmp","compare results of offline and online "
                              "searches",&arguments->docompare,false);
  gt_option_parser_add_option(op, option);
  gt_option_is_development_option(option);

  option = gt_option_new_verbose(&arguments->verbose);
  gt_option_parser_add_option(op, option);

  return op;
}
Exemplo n.º 9
0
static GtOptionParser* gt_encseq_encode_option_parser_new(void *tool_arguments)
{
    GtOptionParser *op;
    GtOption *option;
    GtEncseqEncodeArguments *arguments =
        (GtEncseqEncodeArguments*) tool_arguments;

    /* init */
    op = gt_option_parser_new("sequence_file [sequence_file "
                              "[sequence_file ...]]",
                              "Encode sequence files (FASTA/FASTQ, GenBank, "
                              "EMBL) efficiently.");

    /* -showstats */
    option = gt_option_new_bool("showstats",
                                "show compression results",
                                &arguments->showstats,
                                false);
    gt_option_parser_add_option(op, option);

    /* -no_esq_header */
    option = gt_option_new_bool("no_esq_header",
                                "omit the header in the .esq file",
                                &arguments->no_esq_header,
                                false);
    gt_option_is_development_option(option);
    gt_option_parser_add_option(op, option);

    /* encoded sequence options */
    arguments->eopts = gt_encseq_options_register_encoding(op,
                       arguments->indexname,
                       NULL);

    /* -v */
    option = gt_option_new_verbose(&arguments->verbose);
    gt_option_parser_add_option(op, option);

    gt_option_parser_set_min_args(op, 1);

    return op;
}
Exemplo n.º 10
0
static GtOptionParser* gt_show_seedext_option_parser_new(void *tool_arguments)
{
  GtShowSeedextArguments *arguments = tool_arguments;
  GtOptionParser *op;
  GtOption *option, *op_ali, *option_filename, *op_relax_polish,
           *op_seed_extend, *op_sortmatches, *op_showeoplist;

  gt_assert(arguments);
  /* init */
  op = gt_option_parser_new("[options] -f <matchfilename>",
                            "Parse output of a seed extension and show/verify "
                            "the alignment.");

  /* -a */
  op_ali = gt_option_new_bool("a",
                              "show alignment",
                              &arguments->show_alignment,
                              false);
  gt_option_parser_add_option(op, op_ali);

  /* -seed-display */
  option = gt_option_new_bool("seed-display",
                              "Display seeds in #-line and by "
                              "character + (instead of |) in middle "
                              "row of alignment column",
                              &arguments->seed_display,
                              false);
  gt_option_parser_add_option(op, option);

  /* -seed-extend */
  op_seed_extend = gt_option_new_bool("seed-extend",
                              "read the seeds from the # seed: -lines and "
                              "extend them; match lines are ignored; "
                              "match coordindates are displayed",
                              &arguments->seed_extend,
                              false);
  gt_option_parser_add_option(op, op_seed_extend);

  /* -relax-polish */
  op_relax_polish = gt_option_new_bool("relax-polish",
                                       "do not force alignments to have "
                                       "polished ends",
                                       &arguments->relax_polish,false);
  gt_option_parser_add_option(op, op_relax_polish);
  gt_option_is_development_option(op_relax_polish);
  gt_option_imply(op_relax_polish, op_ali);

  /* -sort */
  op_sortmatches = gt_option_new_bool("sort","sort matches in ascending order "
                                             "of their end positon on the "
                                             "query",
                                      &arguments->sortmatches,false);
  gt_option_parser_add_option(op, op_sortmatches);

  /* -e */
  op_showeoplist = gt_option_new_bool("e","show list of edit operations",
                                      &arguments->showeoplist,false);
  gt_option_parser_add_option(op, op_showeoplist);

  /* -f */
  option_filename = gt_option_new_filename("f",
                                          "path to file with match coordinates",
                                          arguments->matchfilename);
  gt_option_is_mandatory(option_filename);
  gt_option_parser_add_option(op, option_filename);

  gt_option_exclude(op_seed_extend, op_sortmatches);
  gt_option_exclude(op_relax_polish, op_seed_extend);
  return op;
}
static GtOptionParser* gt_condenser_search_option_parser_new
                                                      (void *tool_arguments)
{
  GtCondenserSearchArguments *arguments = tool_arguments;
  GtOptionParser *op;
  GtOption *option, *score_opt, *ceval_opt, *feval_opt, *blastp_opt,
           *blastn_opt;
  gt_assert(arguments);

  /* init */
  op = gt_option_parser_new("[option ...]",
                            "Perform a BLAST or "
                            "HMMSEARCH on the given compressed database.");

  /* -blastn */
  blastn_opt = gt_option_new_bool("blastn", "perform blastn search",
                                  &arguments->blastn, false);
  /* -blastp */
  blastp_opt = gt_option_new_bool("blastp", "perform blastp search",
                                  &arguments->blastp, false);
  gt_option_exclude(blastn_opt, blastp_opt);
  gt_option_parser_add_option(op, blastn_opt);
  gt_option_parser_add_option(op, blastp_opt);

  /* -score */
  score_opt = gt_option_new_uword("score", "bitscore threshold for BLAST(p) "
                                  "evalue calculation",
                                  &arguments->bitscore, (GtUword) 30);
  gt_option_parser_add_option(op, score_opt);

  /* -ce */
  ceval_opt = gt_option_new_double("ce",
                                   "coarse e value for coarse blast search",
                                   &arguments->ceval, 5.0);
  gt_option_parser_add_option(op, ceval_opt);

  /* -fe */
  feval_opt = gt_option_new_double("fe", "fine e value for fine blast search, "
                                   "defaults to calculated evalue from the "
                                   "given score",
                                   &arguments->feval, GT_UNDEF_DOUBLE);
  gt_option_hide_default(feval_opt);
  gt_option_parser_add_option(op, feval_opt);
  gt_option_exclude(score_opt, ceval_opt);
  gt_option_exclude(score_opt, feval_opt);

   /* -db */
  option = gt_option_new_filename("db", "path of (compressed) fasta database",
                                arguments->dbpath);
  gt_option_is_mandatory(option);
  gt_option_parser_add_option(op, option);

  /* -query */
  option = gt_option_new_filename("query", "path of fasta query file",
                                arguments->querypath);
  gt_option_is_mandatory(option);
  gt_option_parser_add_option(op, option);

  /* -verbose */
  option = gt_option_new_bool("verbose", "verbose output", &arguments->verbose,
                              false);
  gt_option_parser_add_option(op, option);

  /* -blastthreads */
  option = gt_option_new_int_min("blastthreads", "how many threads for blast "
                                 "to use", &arguments->blthreads, 8, 1);
  gt_option_imply_either_2(option, blastn_opt, blastp_opt);
  gt_option_is_development_option(option);
  gt_option_parser_add_option(op, option);

  gt_output_file_info_register_options(arguments->ofi, op, &arguments->outfp);

 return op;
}
Exemplo n.º 12
0
static GtOptionParser* gt_linspace_align_option_parser_new(void *tool_arguments)
{
  GtLinspaceArguments *arguments = tool_arguments;
  GtOptionParser *op;
  GtOption *optionstrings, *optionfiles, *optionglobal, *optionlocal,
           *optiondna, *optionprotein, *optioncostmatrix, *optionlinearcosts,
           *optionaffinecosts, *optionoutputfile, *optionshowscore,
           *optionshowsequences, *optiondiagonal, *optiondiagonalbonds,
           *optionsimilarity, *optiontsfactor, *optionspacetime,
           *optionscoreonly, *optionwildcardsymbol;

  gt_assert(arguments);

  /* init */
  op = gt_option_parser_new("[ss|ff] sequence1 sequence2 [dna|protein] "
                            "[global|local] [a|l] costs/scores "
                            "[additional options]",
                            "Apply function to compute alignment.");
  gt_option_parser_set_mail_address(op,
                          "<*****@*****.**>");

  /* -bool */
  optionglobal = gt_option_new_bool("global", "global alignment",
                                    &arguments->global, false);
  gt_option_parser_add_option(op, optionglobal);

  optionlocal = gt_option_new_bool("local", "local alignment",
                                   &arguments->local, false);
  gt_option_parser_add_option(op, optionlocal);

  optiondiagonal = gt_option_new_bool("d", "diagonalband alignment",
                                      &arguments->diagonal, false);
  gt_option_parser_add_option(op, optiondiagonal);

  optiondna = gt_option_new_bool("dna", "type of sequences: DNA",
                                 &arguments->dna, false);
  gt_option_parser_add_option(op, optiondna);

  optionprotein = gt_option_new_bool("protein", "type of sequences: protein",
                                      &arguments->protein, false);
  gt_option_parser_add_option(op, optionprotein);

  optionwildcardsymbol = gt_option_new_bool("wildcard", "show symbol used to "
                                            "represented wildcards in output",
                                            &arguments->wildcardshow, false);
  gt_option_parser_add_option(op, optionwildcardsymbol);

  /* special case, if given matrix includes cost values in place of scores */
  optioncostmatrix = gt_option_new_bool("costmatrix", "describes type of "
                                        "given substituation matrix",
                                        &arguments->has_costmatrix, false);
  gt_option_parser_add_option(op, optioncostmatrix);

  optionshowscore = gt_option_new_bool("showscore", "show score for alignment, "
                                       "please note it will calculate costs "
                                       "for global alignments and scores for "
                                       "local alignemnts always, independtly "
                                       "of input ",
                                       &arguments->showscore, false);
  gt_option_parser_add_option(op, optionshowscore);

  optionshowsequences = gt_option_new_bool("showsequences", "show sequences u "
                                           "and v in front of alignment",
                                       &arguments->showsequences, false);
  gt_option_parser_add_option(op, optionshowsequences);

  optionscoreonly = gt_option_new_bool("showonlyscore", "show only score for "
                                       "generated alignment to compare with "
                                       "other algorithms",
                                       &arguments->scoreonly, false);
  gt_option_parser_add_option(op, optionscoreonly);

  optionspacetime = gt_option_new_bool("spacetime", "write space peak and time"
                                       " overall on stdout",
                                       &arguments->spacetime, false);
  gt_option_parser_add_option(op, optionspacetime);

  /* -str */
  optionstrings = gt_option_new_string_array("ss", "input, use two strings",
                                             arguments->strings);
  gt_option_parser_add_option(op, optionstrings);

  optionfiles = gt_option_new_filename_array("ff", "input, use two files",
                                             arguments->files);
  gt_option_parser_add_option(op, optionfiles);

  optionlinearcosts = gt_option_new_string_array("l", "lineargapcosts, "
                                                 "use match, mismatch and "
                                                 "gapcost, alternatively "
                                                 "substituationmatrix and "
                                                 "gapcost",
                                                 arguments->linearcosts);
  gt_option_parser_add_option(op, optionlinearcosts);

  optionaffinecosts = gt_option_new_string_array("a", "affinegapcosts, "
                                           "use match, mismatch, gap_extension "
                                           "and gap_opening, alternatively "
                                           "substituationmatrix, gap_extension "
                                           "and gap_opening",
                                           arguments->affinecosts);
  gt_option_parser_add_option(op, optionaffinecosts);

  optiondiagonalbonds = gt_option_new_string_array("lr", "specified left and "
                                                   "right shift of diagonal",
                                                   arguments->diagonalbonds);
  gt_option_parser_add_option(op, optiondiagonalbonds);

  optionoutputfile = gt_option_new_string("o", "print alignment, "
                                          "use outputfile",
                                          arguments->outputfile, "stdout");
  gt_option_parser_add_option(op, optionoutputfile);

  /* -ulong */
  optiontsfactor = gt_option_new_ulong("t", "timesquarefactor to organize "
                                       "time and space",
                                       &arguments->timesquarefactor,1);
  gt_option_parser_add_option(op, optiontsfactor);

  /* -double */
  optionsimilarity = gt_option_new_probability("similarity", "specified left "
                                               "and right shift of diagonal by "
                                               "similarity of sequences, "
                                               "0 <= similarty <= 1",
                                               &arguments->similarity, 0);
  gt_option_parser_add_option(op, optionsimilarity);

  /* dependencies */
  gt_option_is_mandatory_either(optionstrings, optionfiles);
  gt_option_is_mandatory_either(optiondna, optionprotein);
  gt_option_exclude(optionlocal, optionglobal);
  gt_option_exclude(optionlinearcosts, optionaffinecosts);
  gt_option_exclude(optiondna, optionprotein);
  gt_option_exclude(optionshowsequences, optionscoreonly);
  gt_option_exclude(optionsimilarity, optiondiagonalbonds);
  gt_option_imply_either_2(optionfiles, optionglobal, optionlocal);
  gt_option_imply_either_2(optiondna, optionstrings, optionfiles);
  gt_option_imply_either_2(optionstrings, optionglobal, optionlocal);
  gt_option_imply_either_2(optionprotein, optionstrings, optionfiles);
  gt_option_imply_either_2(optionlocal, optionlinearcosts, optionaffinecosts);
  gt_option_imply_either_2(optionglobal, optionlinearcosts, optionaffinecosts);
  gt_option_imply_either_2(optionshowscore,optionlinearcosts,optionaffinecosts);
  gt_option_imply_either_2(optionshowsequences, optionstrings, optionfiles);
  gt_option_imply_either_2(optionscoreonly,optionlinearcosts,optionaffinecosts);
  gt_option_imply(optiondiagonal, optionglobal);
  gt_option_imply(optiondiagonalbonds, optiondiagonal);
  gt_option_imply(optionsimilarity, optiondiagonal);
  gt_option_imply(optioncostmatrix, optionprotein);

  /* extended options */
  gt_option_is_extended_option(optiontsfactor);
  gt_option_is_extended_option(optionshowscore);
  gt_option_is_extended_option(optionwildcardsymbol);
  gt_option_is_extended_option(optioncostmatrix);

  /* development option(s) */
  gt_option_is_development_option(optionspacetime);
  gt_option_is_development_option(optionscoreonly);/*only useful to test*/

  return op;
}
Exemplo n.º 13
0
static GtOptionParser* gt_readjoiner_assembly_option_parser_new(
    void *tool_arguments)
{
  GtReadjoinerAssemblyArguments *arguments = tool_arguments;
  GtOptionParser *op;
  GtOption *option, *errors_option, *deadend_option, *v_option,
           *q_option, *bubble_option, *deadend_depth_option;
  gt_assert(arguments);

  /* init */
  op = gt_option_parser_new("[option ...]",
      "Construct string graph and output contigs.");

  /* -readset */
  option = gt_option_new_string("readset", "specify the readset name",
      arguments->readset, NULL);
  gt_option_parser_add_option(op, option);
  gt_option_is_mandatory(option);

  /* -spmfiles */
  option = gt_option_new_uint_min("spmfiles", "number of SPM files to read\n"
      "this must be equal to the value of -j for the overlap phase",
      &arguments->nspmfiles, 1U, 1U);
  gt_option_is_extended_option(option);
  gt_option_parser_add_option(op, option);

  /* -l */
  option = gt_option_new_uint_min("l", "specify the minimum SPM length",
      &arguments->minmatchlength, 0, 2U);
  gt_option_is_extended_option(option);
  gt_option_parser_add_option(op, option);

  /* -depthcutoff */
  option = gt_option_new_uint_min("depthcutoff", "specify the minimal "
      "number of nodes in a contig",
      &arguments->depthcutoff, 3U, 1U);
  gt_option_is_extended_option(option);
  gt_option_parser_add_option(op, option);

  /* -lengthcutoff */
  option = gt_option_new_uint_min("lengthcutoff", "specify the minimal "
      "length of a contig",
      &arguments->lengthcutoff, 100U, 1U);
  gt_option_is_extended_option(option);
  gt_option_parser_add_option(op, option);

  /* -redtrans */
  option = gt_option_new_bool("redtrans", "reduce transitive edges",
      &arguments->redtrans, false);
  gt_option_is_development_option(option);
  gt_option_parser_add_option(op, option);

  /* -errors */
  errors_option = gt_option_new_bool("errors", "search graph features which "
      "may originate from sequencing errors and remove them",
      &arguments->errors, false);
  gt_option_is_extended_option(errors_option);
  gt_option_parser_add_option(op, errors_option);

  /* -bubble */
  bubble_option = gt_option_new_uint("bubble", "number of rounds of p-bubble "
      "removal to perform", &arguments->bubble, 3U);
  gt_option_is_extended_option(bubble_option);
  gt_option_imply(bubble_option, errors_option);
  gt_option_parser_add_option(op, bubble_option);

  /* -deadend */
  deadend_option = gt_option_new_uint("deadend", "number of rounds of "
      "dead end removal to perform a dead end",
      &arguments->deadend, 10U);
  gt_option_is_extended_option(deadend_option);
  gt_option_imply(deadend_option, errors_option);
  gt_option_parser_add_option(op, deadend_option);

  /* -deadend-depth */
  deadend_depth_option = gt_option_new_uint_min("deadend-depth", "specify the "
      "maximal depth of a path to an end-vertex by which the path shall be "
      "considered a dead end",
      &arguments->deadend_depth, 10U, 1U);
  gt_option_is_extended_option(deadend_depth_option);
  gt_option_imply(deadend_depth_option, errors_option);
  gt_option_parser_add_option(op, deadend_depth_option);

  /* -paths2seq */
  option = gt_option_new_bool("paths2seq", "read <indexname>"
      GT_READJOINER_SUFFIX_CONTIG_PATHS " and write "
      "<indexname>" GT_READJOINER_SUFFIX_CONTIGS,
      &arguments->paths2seq, false);
  gt_option_is_development_option(option);
  gt_option_parser_add_option(op, option);

  /* -buffersize */
  option = gt_option_new_string("buffersize", "specify size for read buffer"
      " of paths2seq phase (in bytes, the keywords 'MB' and 'GB' are allowed)",
                       arguments->buffersizearg, NULL);
  gt_option_is_development_option(option);
  gt_option_parser_add_option(op, option);
  arguments->refoptionbuffersize = gt_option_ref(option);

  /* -vd */
  option = gt_option_new_bool("vd", "use verbose descriptions for contigs",
      &arguments->vd, false);
  gt_option_is_development_option(option);
  gt_option_parser_add_option(op, option);

  /* -astat */
  option = gt_option_new_bool("astat", "calculate A-statistics for each contig",
      &arguments->astat, false);
  gt_option_is_development_option(option);
  gt_option_parser_add_option(op, option);

  /* -cov */
  option = gt_option_new_double("cov", "average coverage value to use for the "
      "A-statistics calculation", &arguments->coverage, (double)0);
  gt_option_is_development_option(option);
  gt_option_parser_add_option(op, option);

  /* -copynum */
  option = gt_option_new_bool("copynum", "load reads copy numbers list from "
      "file for the A-statistics calculation",
      &arguments->copynum, false);
  gt_option_is_development_option(option);
  gt_option_parser_add_option(op, option);

  /* -v */
  v_option = gt_option_new_verbose(&arguments->verbose);
  gt_option_parser_add_option(op, v_option);

  /* -q */
  q_option = gt_option_new_bool("q", "suppress standard output messages",
      &arguments->quiet, false);
  gt_option_parser_add_option(op, q_option);
  gt_option_exclude(q_option, v_option);

  /* -load */
  option = gt_option_new_bool("load", "save the string graph from file",
      &arguments->load, false);
  gt_option_is_development_option(option);
  gt_option_parser_add_option(op, option);

  /* -save */
  option = gt_option_new_bool("save", "save the string graph to file",
      &arguments->save, false);
  gt_option_is_development_option(option);
  gt_option_parser_add_option(op, option);

  /* -show_contigs_info */
  option = gt_option_new_bool("cinfo", "output additional files required "
      "for contigs graph construction (eqlen only)",
      &arguments->show_contigs_info, false);
  gt_option_is_development_option(option);
  gt_option_parser_add_option(op, option);

  gt_option_parser_set_version_func(op, gt_readjoiner_show_version);
  gt_option_parser_set_max_args(op, 0);

  return op;
}
Exemplo n.º 14
0
static GtOptionParser* gt_genomediff_option_parser_new(void *tool_arguments)
{
    GtGenomediffArguments *arguments = tool_arguments;
    GtOptionParser *op;
    GtOption *option, *option_unitfile;
    static const char *indextypes[] = { "esa", "pck", "encseq", NULL };

    gt_assert(arguments);

    /* init */
    op = gt_option_parser_new("[option ...] "
                              "(INDEX | -indexname NAME SEQFILE SEQFILE [...]) ",
                              "Calculates Kr: pairwise distances between genomes.");

    /* options */
    option = gt_option_new_choice("indextype", "specify type of index, one of: "
                                  "esa|pck|encseq. Where encseq is an encoded "
                                  "sequence and an enhanced suffix array will be "
                                  "constructed only in memory.",
                                  arguments->indextype, indextypes[2],
                                  indextypes);
    gt_option_parser_add_option(op, option);

    option = gt_option_new_string("indexname", "Basename of encseq to construct.",
                                  arguments->indexname, NULL);
    gt_option_parser_add_option(op, option);

    /*-unitfile*/
    option_unitfile =
        gt_option_new_filename("unitfile",
                               "specifies genomic units, "
                               "see below for description.",
                               arguments->unitfile);
    gt_option_parser_add_option(op, option_unitfile);
    arguments->ref_unitfile = gt_option_ref(option_unitfile);

    /* encseq options */
    arguments->loadopts =
        gt_encseq_options_register_loading(op, arguments->indexname);

    gt_option_is_development_option(
        gt_encseq_options_lossless_option(arguments->loadopts));
    /* esa options */
    arguments->idxopts =
        gt_index_options_register_esa_noout(op);
    gt_option_is_development_option(
        gt_index_options_spmopt_option(arguments->idxopts));

    /* scan */
    option = gt_option_new_bool("scan", "do not load esa index but scan "
                                "it sequentially.", &arguments->scanfile, true);
    gt_option_is_extended_option(option);
    gt_option_parser_add_option(op, option);

    /* dev options */
    /* -max_n */
    option = gt_option_new_uword("max_n", "Number of precalculated values "
                                 "for ln(n!) and pmax(x).",
                                 &arguments->max_ln_n_fac, 1000UL);
    gt_option_is_development_option(option);
    gt_option_parser_add_option(op, option);

    /* -maxdepth */
    option =  gt_option_new_int("maxdepth", "max depth of .pbi-file, use with "
                                "-indextype pck.",
                                &arguments->user_max_depth, -1);
    gt_option_is_development_option(option);
    gt_option_parser_add_option(op, option);

    /* thresholds */
    /* divergence error */
    option = gt_option_new_double("thr",
                                  "Threshold for difference (du, dl) in "
                                  "divergence calculation.\n"
                                  "default: 1e-9",
                                  &arguments->divergence_threshold,
                                  1e-9);
    gt_option_is_extended_option(option);
    gt_option_hide_default(option);
    gt_option_parser_add_option(op, option);

    /* expected shulen error */
    option = gt_option_new_double("abs_err",
                                  "absolute error for expected shulen "
                                  "calculation.\n"
                                  "default: 1e-5",
                                  &arguments->divergence_abs_err,
                                  1e-5);
    gt_option_is_extended_option(option);
    gt_option_hide_default(option);
    gt_option_parser_add_option(op, option);

    /* relative expected shulen error */
    option = gt_option_new_double("rel_err",
                                  "relative error for expected shulen "
                                  "calculation.\n"
                                  "default: 1e-3",
                                  &arguments->divergence_rel_err,
                                  1e-3);
    gt_option_is_extended_option(option);
    gt_option_hide_default(option);
    gt_option_parser_add_option(op, option);

    /* M */
    option = gt_option_new_double("M",
                                  "threshold for minimum logarithm.\n"
                                  "default: DBL_MIN",
                                  &arguments->divergence_m,
                                  DBL_MIN);
    gt_option_is_extended_option(option);
    gt_option_hide_default(option);
    gt_option_parser_add_option(op, option);

    /* -v */
    option = gt_option_new_verbose(&arguments->verbose);
    gt_option_parser_add_option(op, option);

    /* mail */
    gt_option_parser_set_mail_address(op, "<*****@*****.**>");
    /* doc */
    gt_option_parser_set_comment_func(op, gt_gtdata_show_help, NULL);
    return op;
}
Exemplo n.º 15
0
static GtOptionParser* gt_ltrdigest_option_parser_new(void *tool_arguments)
{
  GtLTRdigestOptions *arguments = tool_arguments;
  GtOptionParser *op;
  GtOption *o, *ot, *oto;
  GtOption *oh, *oc, *oeval;
  static const char *cutoffs[] = {"NONE", "GA", "TC", NULL};
  static GtRange pptlen_defaults           = { 8UL, 30UL},
                 uboxlen_defaults          = { 3UL, 30UL},
                 pbsalilen_defaults        = {11UL, 30UL},
                 pbsoffsetlen_defaults     = { 0UL,  5UL},
                 pbstrnaoffsetlen_defaults = { 0UL,  5UL};
  gt_assert(arguments);

  /* init */
  op = gt_option_parser_new("[option ...] gff3_file",
                            "Identifies and annotates sequence features in LTR "
                            "retrotransposon candidates.");

  /* Output files */

  oto = gt_option_new_string("outfileprefix",
                             "prefix for output files (e.g. 'foo' will create "
                             "files called 'foo_*.csv' and 'foo_*.fas')\n"
                             "Omit this option for GFF3 output only.",
                             arguments->prefix,
                             NULL);
  gt_option_parser_add_option(op, oto);
  gt_option_hide_default(oto);

  o = gt_option_new_bool("metadata",
                         "output metadata (run conditions) to separate file",
                         &arguments->print_metadata,
                         true);
  gt_option_parser_add_option(op, o);
  gt_option_imply(o, oto);

  o = gt_option_new_uint("seqnamelen",
                         "set maximal length of sequence names in FASTA headers"
                         " (e.g. for clustalw or similar tools)",
                         &arguments->seqnamelen,
                         20U);
  gt_option_parser_add_option(op, o);

  /* PPT search options */

  o = gt_option_new_range("pptlen",
                          "required PPT length range",
                          &arguments->ppt_len,
                          &pptlen_defaults);
  gt_option_parser_add_option(op, o);

  o = gt_option_new_range("uboxlen",
                          "required U-box length range",
                          &arguments->ubox_len,
                          &uboxlen_defaults);
  gt_option_parser_add_option(op, o);

  o = gt_option_new_uint("uboxdist",
                         "allowed U-box distance range from PPT",
                         &arguments->max_ubox_dist, 0);
  gt_option_parser_add_option(op, o);

  o = gt_option_new_uint("pptradius",
                         "radius around beginning of 3' LTR "
                         "to search for PPT",
                         &arguments->ppt_radius,
                         30U);
  gt_option_parser_add_option(op, o);

  o = gt_option_new_probability("pptrprob",
                                "purine emission probability inside PPT",
                                &arguments->ppt_purine_prob,
                                PPT_PURINE_PROB);
  gt_option_parser_add_option(op, o);
  gt_option_is_extended_option(o);

  o = gt_option_new_probability("pptyprob",
                                "pyrimidine emission probability inside PPT",
                                &arguments->ppt_pyrimidine_prob,
                                PPT_PYRIMIDINE_PROB);
  gt_option_parser_add_option(op, o);
  gt_option_is_extended_option(o);

  o = gt_option_new_probability("pptgprob",
                                "background G emission probability outside PPT",
                                &arguments->bkg_g_prob,
                                BKG_G_PROB);
  gt_option_parser_add_option(op, o);
  gt_option_is_extended_option(o);

  o = gt_option_new_probability("pptcprob",
                                "background C emission probability outside PPT",
                                &arguments->bkg_c_prob,
                                BKG_C_PROB);
  gt_option_parser_add_option(op, o);
  gt_option_is_extended_option(o);

  o = gt_option_new_probability("pptaprob",
                                "background A emission probability outside PPT",
                                &arguments->bkg_a_prob,
                                BKG_A_PROB);
  gt_option_parser_add_option(op, o);
  gt_option_is_extended_option(o);

  o = gt_option_new_probability("ppttprob",
                                "background T emission probability outside PPT",
                                &arguments->bkg_t_prob,
                                BKG_T_PROB);
  gt_option_parser_add_option(op, o);
  gt_option_is_extended_option(o);

  o = gt_option_new_probability("pptuprob",
                                "U/T emission probability inside U-box",
                                &arguments->ubox_u_prob,
                                UBOX_U_PROB);
  gt_option_parser_add_option(op, o);
  gt_option_is_extended_option(o);

  /* PBS search options */

  ot = gt_option_new_filename("trnas",
                              "tRNA library in multiple FASTA format for PBS "
                              "detection\n"
                              "Omit this option to disable PBS search.",
                              arguments->trna_lib);
  gt_option_parser_add_option(op, ot);
  gt_option_hide_default(ot);

  o = gt_option_new_range("pbsalilen",
                          "required PBS/tRNA alignment length range",
                          &arguments->alilen,
                          &pbsalilen_defaults);
  gt_option_parser_add_option(op, o);
  gt_option_imply(o, ot);

  o = gt_option_new_range("pbsoffset",
                          "allowed PBS offset from LTR boundary range",
                          &arguments->offsetlen,
                          &pbsoffsetlen_defaults);
  gt_option_parser_add_option(op, o);
  gt_option_imply(o, ot);

  o = gt_option_new_range("pbstrnaoffset",
                          "allowed PBS/tRNA 3' end alignment offset range",
                          &arguments->trnaoffsetlen,
                          &pbstrnaoffsetlen_defaults);
  gt_option_parser_add_option(op, o);
  gt_option_imply(o, ot);

  o = gt_option_new_uint("pbsmaxedist",
                         "maximal allowed PBS/tRNA alignment unit "
                         "edit distance",
                         &arguments->max_edist,
                         1U);
  gt_option_parser_add_option(op, o);
  gt_option_imply(o, ot);

  o = gt_option_new_uint("pbsradius",
                         "radius around end of 5' LTR "
                         "to search for PBS",
                         &arguments->pbs_radius,
                         30U);
  gt_option_parser_add_option(op, o);
  gt_option_imply(o, ot);

 /* Protein domain search options */

  oh = gt_option_new_filename_array("hmms",
                                    "profile HMM models for domain detection "
                                    "(separate by spaces, finish with --) in "
                                    "HMMER3 format\n"
                                    "Omit this option to disable pHMM search.",
                                    arguments->hmm_files);
  gt_option_parser_add_option(op, oh);

  oeval = gt_option_new_probability("pdomevalcutoff",
                                    "global E-value cutoff for pHMM search\n"
                                    "default 1E-6",
                                    &arguments->evalue_cutoff,
                                    0.000001);
  gt_option_parser_add_option(op, oeval);
  gt_option_is_extended_option(oeval);
  gt_option_hide_default(oeval);
  gt_option_imply(oeval, oh);

  oc = gt_option_new_choice("pdomcutoff", "model-specific score cutoff\n"
                                       "choose from TC (trusted cutoff) | "
                                       "GA (gathering cutoff) | "
                                       "NONE (no cutoffs)",
                             arguments->cutoffs, cutoffs[0], cutoffs);
  gt_option_parser_add_option(op, oc);
  gt_option_is_extended_option(oeval);
  gt_option_imply(oeval, oh);

  o = gt_option_new_bool("aliout",
                         "output pHMM to amino acid sequence alignments",
                         &arguments->write_alignments,
                         false);
  gt_option_parser_add_option(op, o);
  gt_option_imply(o, oh);
  gt_option_imply(o, oto);

  o = gt_option_new_bool("aaout",
                         "output amino acid sequences for protein domain "
                         "hits",
                         &arguments->write_aaseqs,
                         false);
  gt_option_parser_add_option(op, o);
  gt_option_imply(o, oh);
  gt_option_imply(o, oto);

  o = gt_option_new_bool("allchains",
                           "output features from all chains and unchained "
                           "features, labeled with chain numbers",
                           &arguments->output_all_chains,
                           false);
  gt_option_parser_add_option(op, o);
  gt_option_imply(o, oh);

  o = gt_option_new_uint("maxgaplen",
                         "maximal allowed gap size between fragments (in amino "
                         "acids) when chaining pHMM hits for a protein domain",
                         &arguments->chain_max_gap_length,
                         50U);
  gt_option_parser_add_option(op, o);
  gt_option_is_extended_option(o);
  gt_option_imply(o, oh);

  o = gt_option_new_uword("threads",
                          "DEPRECATED, only included for compatibility reasons!"
                          " Use the -j parameter of the 'gt' call instead.",
                          &arguments->nthreads,
                          0);
  gt_option_parser_add_option(op, o);
  gt_option_is_development_option(o);

  /* Extended PBS options */

  o = gt_option_new_int("pbsmatchscore",
                        "match score for PBS/tRNA alignments",
                        &arguments->ali_score_match,
                        5);
  gt_option_parser_add_option(op, o);
  gt_option_is_extended_option(o);
  gt_option_imply(o, ot);

  o = gt_option_new_int("pbsmismatchscore",
                        "mismatch score for PBS/tRNA alignments",
                        &arguments->ali_score_mismatch,
                        -10);
  gt_option_parser_add_option(op, o);
  gt_option_is_extended_option(o);
  gt_option_imply(o, ot);

  o = gt_option_new_int("pbsinsertionscore",
                        "insertion score for PBS/tRNA alignments",
                        &arguments->ali_score_insertion,
                        -20);
  gt_option_parser_add_option(op, o);
  gt_option_is_extended_option(o);
  gt_option_imply(o, ot);

  o = gt_option_new_int("pbsdeletionscore",
                        "deletion score for PBS/tRNA alignments",
                        &arguments->ali_score_deletion,
                        -20);
  gt_option_parser_add_option(op, o);
  gt_option_is_extended_option(o);
  gt_option_imply(o, ot);

  /* verbosity */

  o = gt_option_new_verbose(&arguments->verbose);
  gt_option_parser_add_option(op, o);

  /* output file options */

  gt_output_file_info_register_options(arguments->ofi, op, &arguments->outfp);

  /* region mapping and sequence source options */

  gt_seqid2file_register_options_ext(op, arguments->s2fi, false, false);

  return op;
}
Exemplo n.º 16
0
static GtOptionParser* gt_select_option_parser_new(void *tool_arguments)
{
  SelectArguments *arguments = tool_arguments;
  GtOptionParser *op;
  GtOption *option, *contain_option, *overlap_option, *minaveragessp_option,
           *singleintron_option, *optiondroppedfile;
  gt_assert(arguments);

  static const char *filter_logic[] = {
    "AND",
    "OR",
    NULL
  };

  /* init */
  op = gt_option_parser_new("[option ...] [GFF3_file ...]",
                            "Select certain features (specified by the used "
                            "options) from given GFF3 file(s).");

  /* -seqid */
  option = gt_option_new_string("seqid", "select feature with the given "
                                "sequence ID (all comments are selected). ",
                                arguments->seqid, NULL);
  gt_option_parser_add_option(op, option);

  /* -source */
  option = gt_option_new_string("source", "select feature with the given "
                                "source (the source is column 2 in regular "
                                "GFF3 lines)" , arguments->source, NULL);
  gt_option_parser_add_option(op, option);

  /* -contain */
  contain_option = gt_option_new_range("contain", "select all features which "
                                       "are contained in the given range",
                                       &arguments->contain_range, NULL);
  gt_option_parser_add_option(op, contain_option);

  /* -overlap */
  overlap_option = gt_option_new_range("overlap", "select all features which "
                                       "do overlap with the given range",
                                       &arguments->overlap_range, NULL);
  gt_option_parser_add_option(op, overlap_option);

  /* -strand */
  option = gt_option_new_string(GT_STRAND_OPT, "select all top-level features"
                                "(i.e., features without parents) whose strand "
                                "equals the given one (must be one of '"
                                GT_STRAND_CHARS"')", arguments->gt_strand_char,
                                NULL);
  gt_option_parser_add_option(op, option);

  /* -targetstrand */
  option = gt_option_new_string(TARGETGT_STRAND_OPT, "select all top-level "
                                "features (i.e., features without parents) "
                                "which have exactly one target attribute whose "
                                "strand equals the given one (must be one of '"
                                GT_STRAND_CHARS"')",
                                arguments->targetgt_strand_char, NULL);
  gt_option_parser_add_option(op, option);

  /* -targetbest */
  option = gt_option_new_bool("targetbest", "if multiple top-level features "
                             "(i.e., features without parents) with exactly "
                             "one target attribute have the same target_id, "
                             "keep only the feature with the best score. If "
                             "-"TARGETGT_STRAND_OPT" is used at the same time, "
                             "this option is applied after "
                             "-"TARGETGT_STRAND_OPT".\n"
                             "Memory consumption is proportional to the input "
                             "file size(s).", &arguments->targetbest, false);
  gt_option_parser_add_option(op, option);

  /* -hascds */
  option = gt_option_new_bool("hascds", "select all top-level features which "
                              "do have a CDS child", &arguments->has_CDS,
                              false);
  gt_option_parser_add_option(op, option);

  /* -maxgenelength */
  option = gt_option_new_uword_min("maxgenelength", "select genes up to the "
                                   "given maximum length",
                                   &arguments->max_gene_length, GT_UNDEF_UWORD,
                                   1);
  gt_option_parser_add_option(op, option);

  /* -maxgenenum */
  option = gt_option_new_uword("maxgenenum", "select the first genes up to the "
                               "given maximum number", &arguments->max_gene_num,
                               GT_UNDEF_UWORD);
  gt_option_parser_add_option(op, option);

  /* -mingenescore */
  option = gt_option_new_double("mingenescore", "select genes with the given "
                                "minimum score", &arguments->min_gene_score,
                                GT_UNDEF_DOUBLE);
  gt_option_parser_add_option(op, option);

  /* -maxgenescore */
  option = gt_option_new_double("maxgenescore", "select genes with the given "
                                "maximum score", &arguments->max_gene_score,
                                GT_UNDEF_DOUBLE);
  gt_option_parser_add_option(op, option);

  /* -minaveragessp */
  minaveragessp_option =
    gt_option_new_probability("minaveragessp",
                              "set the minimum average splice site probability",
                              &arguments->min_average_splice_site_prob,
                              GT_UNDEF_DOUBLE);
  gt_option_parser_add_option(op, minaveragessp_option);

  /* -singleintronfactor */
  singleintron_option =
    gt_option_new_double_min("singleintronfactor",
                             "factor to multiplicate the average splice site "
                             "probability with for single introns before "
                             "comparing it to the minimum average splice site "
                             "probability", &arguments->single_intron_factor,
                             1.0, 1.0);
  gt_option_is_development_option(singleintron_option);
  gt_option_parser_add_option(op, singleintron_option);

  /* -featurenum */
  option = gt_option_new_uword_min("featurenum",
                                   "select feature tree occurring "
                                   "at given position in input",
                                   &arguments->feature_num, GT_UNDEF_UWORD, 1);
  gt_option_is_development_option(option);
  gt_option_parser_add_option(op, option);

  /* -filter_files */
  option = gt_option_new_filename_array("rule_files",
                                        "specify Lua files to be used "
                                        "for selection",
                                        arguments->filter_files);
  gt_option_parser_add_option(op, option);

  /* -filter_logic */
  option = gt_option_new_choice("rule_logic", "select how multiple Lua "
                                "files should be combined\nchoose from AND|OR",
                                arguments->filter_logic, filter_logic[0],
                                filter_logic);
  gt_option_parser_add_option(op, option);

  /* -nh_file */
  optiondroppedfile = gt_option_new_filename("dropped_file",
                                             "save non-selected features to "
                                             "file",
                                             arguments->dropped_file);
  gt_option_parser_add_option(op, optiondroppedfile);

  /* -v */
  option = gt_option_new_verbose(&arguments->verbose);
  gt_option_parser_add_option(op, option);

  /* option exclusions */
  gt_option_exclude(contain_option, overlap_option);

  /* option implications */
  gt_option_imply(singleintron_option, minaveragessp_option);

  /* output file options */
  gt_output_file_info_register_options(arguments->ofi, op, &arguments->outfp);

  gt_option_parser_set_comment_func(op, gt_gtdata_show_help, NULL);

  return op;
}
Exemplo n.º 17
0
static GtOptionParser* gt_matstat_option_parser_new(void *tool_arguments)
{
  Gfmsubcallinfo *arguments = tool_arguments;
  GtOptionParser *op;
  gt_assert(arguments);

  op = gt_option_parser_new("[options ...] -query queryfile [...]",
                         arguments->doms
                         ? "Compute matching statistics."
                         : "Compute length of minimum unique prefixes.");
  gt_option_parser_set_mail_address(op,"<*****@*****.**>");

  arguments->optionfmindex = gt_option_new_string("fmi", "specify fmindex",
                                                  arguments->indexname,NULL);
  gt_option_parser_add_option(op, arguments->optionfmindex);

  arguments->optionesaindex = gt_option_new_string("esa",
                                                   "specify suffix array",
                                                   arguments->indexname,NULL);
  gt_option_parser_add_option(op, arguments->optionesaindex);

  arguments->optionpckindex = gt_option_new_string("pck",
                                                   "specify packed index",
                                                   arguments->indexname,NULL);
  gt_option_parser_add_option(op, arguments->optionpckindex);

  gt_option_exclude(arguments->optionfmindex,arguments->optionesaindex);
  gt_option_exclude(arguments->optionpckindex,arguments->optionesaindex);
  gt_option_exclude(arguments->optionpckindex,arguments->optionfmindex);

  arguments->optionquery = gt_option_new_filename_array("query",
                                                     "specify queryfiles",
                                                     arguments->queryfilenames);
  gt_option_is_mandatory(arguments->optionquery);
  gt_option_parser_add_option(op, arguments->optionquery);

  arguments->optionmin = gt_option_new_uword_min("min",
                                   "only output length "
                                   "if >= given minimum length",
                                   &arguments->minlength.
                                          valueunsignedlong,
                                   0,(GtUword) 1);
  gt_option_parser_add_option(op, arguments->optionmin);

  arguments->optionmax = gt_option_new_uword_min("max",
                                   "only output length "
                                   "if <= given maximum length",
                                   &arguments->maxlength.
                                          valueunsignedlong,
                                   0,(GtUword) 1);
  gt_option_parser_add_option(op, arguments->optionmax);

  arguments->optionoutput = gt_option_new_string_array("output",
                   arguments->doms
                     ? "set output flags (sequence, querypos, subjectpos)"
                     : "set output flags (sequence, querypos)",
                   arguments->flagsoutputoption);
  gt_option_parser_add_option(op, arguments->optionoutput);

  if (arguments->doms) {
    arguments->optionverify = gt_option_new_bool("verify",
                                                 "verify witness positions",
                                                 &arguments->verifywitnesspos,
                                                 false);
    gt_option_is_development_option(arguments->optionverify);
    gt_option_parser_add_option(op, arguments->optionverify);
  } else
  {
    arguments->verifywitnesspos = false;
  }

  gt_option_parser_refer_to_manual(op);

  return op;
}
Exemplo n.º 18
0
static GtOptionParser *gt_tyr_mkindex_option_parser_new(void *tool_arguments)
{
  GtOptionParser *op;
  GtOption *option,
           *optionminocc,
           *optionmaxocc,
           *optionpl,
           *optionstoreindex,
           *optionstorecounts,
           *optionscan,
           *optionesa;
  Tyr_mkindex_options *arguments = tool_arguments;

  op = gt_option_parser_new("[options] -esa suffixerator-index [options]",
                            "Count and index k-mers in the given enhanced "
                            "suffix array for a fixed value of k.");
  gt_option_parser_set_mailaddress(op,"<*****@*****.**>");

  optionesa = gt_option_new_string("esa","specify suffixerator-index\n"
                                   "(mandatory option)",
                                   arguments->str_inputindex,
                                   NULL);
  gt_option_is_mandatory(optionesa);
  gt_option_parser_add_option(op, optionesa);

  option = gt_option_new_ulong("mersize",
                               "Specify the mer size.",
                               &arguments->mersize,
                               20UL);
  gt_option_parser_add_option(op, option);

  optionminocc
    = gt_option_new_ulong("minocc",
                          "Specify the minimum occurrence number for "
                          "the mers to output/index",
                          &arguments->userdefinedminocc,
                          0);
  gt_option_parser_add_option(op, optionminocc);

  optionmaxocc
    = gt_option_new_ulong("maxocc",
                          "Specify the maximum occurrence number for "
                          "the mers to output/index",
                          &arguments->userdefinedmaxocc,
                          0);
  gt_option_parser_add_option(op, optionmaxocc);

  optionpl = gt_option_new_uint_min("pl",
                 "specify prefix length for bucket boundary construction\n"
                 "recommendation: use without argument;\n"
                 "then a reasonable prefix length is automatically determined",
                 &arguments->userdefinedprefixlength,
                 0,
                 1U);
  gt_option_argument_is_optional(optionpl);
  gt_option_parser_add_option(op, optionpl);
  arguments->refoptionpl = gt_option_ref(optionpl);

  optionstoreindex = gt_option_new_string("indexname",
                                          "store the mers specified by options "
                                          "-maxocc and -minocc in an index",
                                          arguments->str_storeindex, NULL);
  gt_option_parser_add_option(op, optionstoreindex);

  optionstorecounts = gt_option_new_bool("counts", "store counts of the mers",
                                         &arguments->storecounts,false);
  gt_option_parser_add_option(op, optionstorecounts);

  option = gt_option_new_bool("test", "perform tests to verify program "
                                      "correctness", &arguments->performtest,
                                      false);
  gt_option_is_development_option(option);
  gt_option_parser_add_option(op, option);

  optionscan = gt_option_new_bool("scan",
                                  "read enhanced suffix array sequentially "
                                  "instead of mapping it to memory",
                                  &arguments->scanfile,
                                  false);
  gt_option_parser_add_option(op, optionscan);

  option = gt_option_new_verbose(&arguments->verbose);
  gt_option_parser_add_option(op, option);

  gt_option_imply(optionpl, optionstoreindex);
  gt_option_imply(optionstorecounts, optionstoreindex);
  gt_option_imply_either_2(optionstoreindex,optionminocc,optionmaxocc);
  return op;
}
Exemplo n.º 19
0
static GtOptionParser *gt_repfind_option_parser_new(void *tool_arguments)
{
  GtOptionParser *op;
  GtOption *option, *reverseoption, *queryoption, *extendseedoption,
           *scanoption, *sampleoption, *forwardoption, *spmoption;
  Maxpairsoptions *arguments = tool_arguments;

  op = gt_option_parser_new("[options] -ii indexname",
                            "Compute maximal repeats.");
  gt_option_parser_set_mail_address(op,"<*****@*****.**>");

  option = gt_option_new_uint_min("l","Specify minimum length of repeats",
                                  &arguments->userdefinedleastlength,
                                  20U,
                                  1U);
  gt_option_parser_add_option(op, option);

  forwardoption = gt_option_new_bool("f","Compute maximal forward repeats",
                                     &arguments->forward,
                                     true);
  gt_option_parser_add_option(op, forwardoption);
  arguments->refforwardoption = gt_option_ref(forwardoption);

  reverseoption = gt_option_new_bool("r","Compute maximal reverse matches",
                                     &arguments->reverse,
                                     false);
  gt_option_parser_add_option(op, reverseoption);

  sampleoption = gt_option_new_uword_min("samples","Specify number of samples",
                                         &arguments->samples,
                                         0,
                                         1UL);
  gt_option_is_development_option(sampleoption);
  gt_option_parser_add_option(op, sampleoption);

  spmoption = gt_option_new_bool("spm","Search for suffix prefix matches",
                                       &arguments->searchspm,
                                       false);
  gt_option_is_development_option(spmoption);
  gt_option_parser_add_option(op, spmoption);

  extendseedoption = gt_option_new_bool("extend","Extend seed to both sides",
                                        &arguments->extendseed,
                                    false);
  gt_option_is_development_option(extendseedoption);
  gt_option_parser_add_option(op, extendseedoption);

  scanoption = gt_option_new_bool("scan","scan index rather than mapping "
                                         "it to main memory",
                                  &arguments->scanfile,
                                  false);
  gt_option_parser_add_option(op, scanoption);

  option = gt_option_new_string("ii",
                                "Specify input index",
                                arguments->indexname, NULL);
  gt_option_parser_add_option(op, option);
  gt_option_is_mandatory(option);

  queryoption = gt_option_new_filename_array("q",
                                             "Specify query files",
                                             arguments->queryfiles);
  gt_option_is_development_option(queryoption);
  gt_option_parser_add_option(op, queryoption);

  option = gt_option_new_bool("v",
                              "be verbose ",
                              &arguments->beverbose,
                              false);
  gt_option_parser_add_option(op, option);

  gt_option_exclude(queryoption,sampleoption);
  gt_option_exclude(queryoption,scanoption);
  gt_option_exclude(queryoption,reverseoption);
  gt_option_exclude(queryoption,spmoption);
  gt_option_exclude(reverseoption,spmoption);
  gt_option_exclude(queryoption,spmoption);
  gt_option_exclude(sampleoption,spmoption);
  return op;
}
Exemplo n.º 20
0
static GtOptionParser* gt_sketch_option_parser_new(void *tool_arguments)
{
  GtSketchArguments *arguments = tool_arguments;
  GtOptionParser *op;
  GtOption *option, *option2;
  static const char *formats[] = { "png",
#ifdef CAIRO_HAS_PDF_SURFACE
    "pdf",
#endif
#ifdef CAIRO_HAS_SVG_SURFACE
    "svg",
#endif
#ifdef CAIRO_HAS_PS_SURFACE
    "ps",
#endif
    NULL
  };
  static const char *inputs[] = {
    "gff",
    "bed",
    "gtf",
    NULL
  };
  gt_assert(arguments);

  /* init */
  op = gt_option_parser_new("[option ...] image_file [GFF3_file ...]",
                            "Create graphical representation of GFF3 "
                            "annotation files.");

  /* -pipe */
  option = gt_option_new_bool("pipe", "use pipe mode (i.e., show all gff3 "
                              "features on stdout)", &arguments->pipe, false);
  gt_option_parser_add_option(op, option);

  /* -flattenfiles */
  option = gt_option_new_bool("flattenfiles", "do not group tracks by source "
                              "file name and remove file names from track "
                              "description", &arguments->flattenfiles, false);
  gt_option_parser_add_option(op, option);

  /* -seqid */
  option = gt_option_new_string("seqid", "sequence region identifier\n"
                                      "default: first one in file",
                            arguments->seqid, NULL);
  gt_option_parser_add_option(op, option);
  gt_option_hide_default(option);

  /* -start */
  option = gt_option_new_uword_min("start", "start position\n"
                                         "default: first region start",
                            &arguments->start, GT_UNDEF_UWORD, 1);
  gt_option_parser_add_option(op, option);
  gt_option_hide_default(option);

  /* -end */
  option2 = gt_option_new_uword("end", "end position\ndefault: last region end",
                            &arguments->end, GT_UNDEF_UWORD);
  gt_option_parser_add_option(op, option2);
  /* -start and -end must be given together */
  gt_option_imply(option, option2);
  gt_option_imply(option2, option);
  gt_option_hide_default(option2);

  /* -width */
  option = gt_option_new_uint_min("width", "target image width (in pixel)",
                                  &arguments->width,
                                  800, 1);
  gt_option_parser_add_option(op, option);

  /* -style */
  option = gt_option_new_string("style", "style file to use",
                                arguments->stylefile,
                                gt_str_get(arguments->stylefile));
  gt_option_parser_add_option(op, option);

  /* -format */
  option = gt_option_new_choice("format", "output graphics format\n"
                                       "choose from png"
#ifdef CAIRO_HAS_PDF_SURFACE
                                       "|pdf"
#endif
#ifdef CAIRO_HAS_SVG_SURFACE
                                       "|svg"
#endif
#ifdef CAIRO_HAS_PS_SURFACE
                                       "|ps"
#endif
                                       "",
                             arguments->format, formats[0], formats);
  gt_option_parser_add_option(op, option);

  /* -input */
  option = gt_option_new_choice("input", "input data format\n"
                                       "choose from gff|bed|gtf",
                             arguments->input, inputs[0], inputs);
  gt_option_parser_add_option(op, option);

  /* -addintrons */
  option = gt_option_new_bool("addintrons", "add intron features between "
                              "existing exon features (before drawing)",
                              &arguments->addintrons, false);
  gt_option_parser_add_option(op, option);

    /* -unsafe */
  option = gt_option_new_bool("unsafe", "enable unsafe mode for style file",
                              &arguments->unsafe, false);
  gt_option_parser_add_option(op, option);

  /* -showrecmaps */
  option = gt_option_new_bool("showrecmaps",
                              "show RecMaps after image creation",
                              &arguments->showrecmaps, false);
  gt_option_is_development_option(option);
  gt_option_parser_add_option(op, option);

  /* -streams */
  option = gt_option_new_bool("streams", "use streams to write data to file",
                              &arguments->use_streams, false);
  gt_option_is_development_option(option);
  gt_option_parser_add_option(op, option);

  /* -v */
  option = gt_option_new_verbose(&arguments->verbose);
  gt_option_parser_add_option(op, option);

  /* -force */
  option = gt_option_new_bool(GT_FORCE_OPT_CSTR, "force writing to output file",
                              &arguments->force, false);
  gt_option_parser_add_option(op, option);

  gt_option_parser_set_min_args(op, 1);

  return op;
}
Exemplo n.º 21
0
static GtOptionParser *gt_idxlocali_option_parser_new (void *tool_arguments)
{
  IdxlocaliOptions *arguments = tool_arguments;
  GtOptionParser *op;
  GtOption *option, *optionesaindex, *optionpckindex, *optiononline, *optioncmp;

  gt_assert (arguments != NULL);
  arguments->indexname = gt_str_new ();
  arguments->queryfiles = gt_str_array_new ();

  op = gt_option_parser_new
    ("[options] -q query-file-names [-esa|-pck] indexname",
     "Find all local alignments using suffix tree.");

  gt_option_parser_set_mailaddress (op, "<*****@*****.**>");
  option = gt_option_new_filenamearray ("q","Specify files containing the "
                                            "query sequences",
                                        arguments->queryfiles);
  gt_option_parser_add_option (op, option);

  option = gt_option_new_long ("match",
                               "Specify match score",
                               &arguments->matchscore, 1L);
  gt_option_parser_add_option (op, option);

  option = gt_option_new_long ("mismatch",
                               "Specify mismatch score",
                               &arguments->mismatchscore, -3L);
  gt_option_parser_add_option (op, option);

  option = gt_option_new_long ("gapstart",
                               "Specify gap start score",
                               &arguments->gapstart, -5L);
  gt_option_parser_add_option (op, option);

  option = gt_option_new_long ("gapextend",
                               "Specify gap extension score",
                               &arguments->gapextend, -2L);
  gt_option_parser_add_option (op, option);

  option = gt_option_new_ulong_min ("th",
                                    "Specify the threshold",
                                     &arguments->threshold, 0, 1UL);
  gt_option_parser_add_option (op, option);
  gt_option_is_mandatory (option);

  optionesaindex = gt_option_new_string ("esa",
                                         "Specify index "
                                         "(enhanced suffix array)",
                                         arguments->indexname, NULL);
  gt_option_parser_add_option (op, optionesaindex);
  arguments->refoptionesaindex = gt_option_ref (optionesaindex);

  optionpckindex = gt_option_new_string ("pck",
                                         "Specify index (packed index)",
                                         arguments->indexname, NULL);
  gt_option_parser_add_option (op, optionpckindex);
  arguments->refoptionpckindex = gt_option_ref (optionpckindex);
  gt_option_exclude (optionesaindex, optionpckindex);
  gt_option_is_mandatory_either (optionesaindex, optionpckindex);

  optiononline = gt_option_new_bool("online","Perform online searches",
                                    &arguments->doonline, false);
  gt_option_parser_add_option(op, optiononline);
  gt_option_is_development_option(optiononline);

  optioncmp = gt_option_new_bool("cmp","Compare results of offline and online "
                                 "searches",
                                 &arguments->docompare, false);
  gt_option_parser_add_option(op,optioncmp);
  gt_option_exclude(optiononline,optioncmp);

  option = gt_option_new_bool("s",
                              "Show alignments",
                              &arguments->showalignment, false);
  gt_option_parser_add_option (op, option);

  option = gt_option_new_verbose(&arguments->verbose);
  gt_option_parser_add_option(op, option);
  return op;
}
Exemplo n.º 22
0
static GtOptionParser* gt_encseq2spm_option_parser_new(void *tool_arguments)
{
    GtEncseq2spmArguments *arguments = tool_arguments;
    GtOptionParser *op;
    GtOption *option, *optionparts, *optionmemlimit;

    gt_assert(arguments);

    /* init */
    op = gt_option_parser_new("[option ...] [file]",
                              "Compute suffix prefix matches "
                              "from encoded sequence.");

    /* -l */
    option = gt_option_new_uint_min("l", "specify the minimum length",
                                    &arguments->minmatchlength, 0, 1U);
    gt_option_parser_add_option(op, option);
    gt_option_is_mandatory(option);

    /* -parts */
    optionparts = gt_option_new_uint("parts", "specify the number of parts",
                                     &arguments->numofparts, 0U);
    gt_option_parser_add_option(op, optionparts);

    /* -memlimit */
    optionmemlimit = gt_option_new_string("memlimit",
                                          "specify maximal amount of memory to be used during "
                                          "index construction (in bytes, the keywords 'MB' "
                                          "and 'GB' are allowed)",
                                          arguments->memlimitarg, NULL);
    gt_option_parser_add_option(op, optionmemlimit);
    gt_option_exclude(optionmemlimit, optionparts);
    arguments->refoptionmemlimit = gt_option_ref(optionmemlimit);

    /* -checksuftab */
    option = gt_option_new_bool("checksuftab", "check the suffix table",
                                &arguments->checksuftab, false);
    gt_option_parser_add_option(op, option);
    gt_option_is_development_option(option);

    /* -singlestrand */
    option = gt_option_new_bool("singlestrand", "use only the forward strand "
                                "of the sequence",
                                &arguments->singlestrand, false);
    gt_option_parser_add_option(op, option);

    /* -spm */
    option = gt_option_new_string("spm", "specify output for spms",
                                  arguments->spmspec, NULL);
    gt_option_parser_add_option(op, option);

    /* -ii */
    option = gt_option_new_string("ii", "specify the input sequence",
                                  arguments->encseqinput, NULL);
    gt_option_parser_add_option(op, option);
    gt_option_is_mandatory(option);

    /* -onlyaccum */
    option = gt_option_new_bool("onlyaccum", "only accumulate codes",
                                &arguments->onlyaccum, false);
    gt_option_parser_add_option(op, option);
    gt_option_is_development_option(option);

    /* -onlyallfirstcodes */
    option = gt_option_new_bool("onlyallfirstcodes", "only determines allcodes",
                                &arguments->onlyallfirstcodes, false);
    gt_option_parser_add_option(op, option);
    gt_option_is_development_option(option);

    /* -addbscachedepth */
    option = gt_option_new_uint("addbscachedepth", "only determines allcodes",
                                &arguments->addbscache_depth, 5U);
    gt_option_parser_add_option(op, option);
    gt_option_is_development_option(option);

    /* -phase2extra */
    option = gt_option_new_string("phase2extra",
                                  "specify  amount of additional space required for "
                                  "the second phase of the computation involving the "
                                  "processing of the intervals (in bytes, "
                                  "the keywords 'MB' and 'GB' are allowed)",
                                  arguments->phase2extraarg, NULL);
    gt_option_parser_add_option(op, option);
    arguments->refoptionphase2extra = gt_option_ref(option);
    gt_option_is_development_option(option);

    /* -radixlarge */
    option = gt_option_new_bool("radixlarge", "use large tables for radixsort",
                                &arguments->radixlarge, false);
    gt_option_parser_add_option(op, option);
    gt_option_is_development_option(option);

    /* -radixparts */
    option = gt_option_new_uint("radixparts", "specify the number of parts "
                                "for radixsort",
                                &arguments->radixparts, 1U);
    gt_option_parser_add_option(op, option);
    gt_option_is_development_option(option);

    /* -singlescan */
    option = gt_option_new_uint("singlescan", "run a single scan: 1=fast; "
                                "2=fast with check; 3=fast with output; "
                                "4=sfx-mapped4-version",
                                &arguments->singlescan, 0);
    gt_option_parser_add_option(op, option);
    gt_option_is_development_option(option);

    /* -forcek */
    option = gt_option_new_uint("forcek", "specify the value of k",
                                &arguments->forcek, 0);
    gt_option_is_development_option(option);
    gt_option_parser_add_option(op, option);

    option = gt_option_new_verbose(&arguments->verbose);
    gt_option_parser_add_option(op, option);

    return op;
}
Exemplo n.º 23
0
static GtIndexOptions*
gt_index_options_register_generic_output(GtOptionParser *op,
                                         GtIndexOptions *idxo,
                                         GtStr *indexname,
                                         GtEncseqOptions *encopts)
{
  gt_assert(idxo != NULL);
  gt_assert(op != NULL && idxo->type != GT_INDEX_OPTIONS_UNDEFINED &&
            encopts != NULL);
  idxo->encopts = encopts;
  idxo->indexname = indexname != NULL ? gt_str_ref(indexname) : NULL;
  idxo->optionkys = gt_option_new_string("kys",
                                   "output/sort according to keys of the form "
                                   "|key| in fasta header",
                                   idxo->kysargumentstring,
                                   "nosort");
  gt_option_argument_is_optional(idxo->optionkys);
  gt_option_imply(idxo->optionkys, gt_encseq_options_sds_option(idxo->encopts));
  gt_option_parser_add_option(op, idxo->optionkys);

  gt_encseq_options_add_readmode_option(op, idxo->dir);

  if (idxo->type == GT_INDEX_OPTIONS_ESA)
  {
    idxo->optionoutsuftab = gt_option_new_bool("suf",
                                   "output suffix array (suftab) to file",
                                   &idxo->outsuftab,
                                   false);
    gt_option_parser_add_option(op, idxo->optionoutsuftab);

    idxo->optionoutlcptab = gt_option_new_bool("lcp",
                                   "output lcp table (lcptab) to file",
                                   &idxo->outlcptab,
                                   false);
    gt_option_parser_add_option(op, idxo->optionoutlcptab);

    idxo->option = gt_option_new_bool("lcpdist",
                              "output distributions of values in lcptab",
                              &idxo->lcpdist,
                              false);
    gt_option_is_extended_option(idxo->option);
    gt_option_imply(idxo->option, idxo->optionoutlcptab);
    gt_option_parser_add_option(op, idxo->option);

    idxo->option = gt_option_new_bool("swallow-tail",
                              "swallow the tail of the suffix array and lcptab",
                              &idxo->swallow_tail,
                              false);
    gt_option_is_development_option(idxo->option);
    gt_option_parser_add_option(op, idxo->option);

    idxo->optionoutbwttab = gt_option_new_bool("bwt",
                                   "output Burrows-Wheeler Transformation "
                                   "(bwttab) to file",
                                   &idxo->outbwttab,
                                   false);
    gt_option_exclude(idxo->optionspmopt, idxo->optionoutbwttab);
    gt_option_parser_add_option(op, idxo->optionoutbwttab);

    idxo->optionoutbcktab = gt_option_new_bool("bck",
                                "output bucket table to file",
                                &idxo->outbcktab,
                                false);
    gt_option_parser_add_option(op, idxo->optionoutbcktab);
  } else {
    idxo->optionoutsuftab
      = idxo->optionoutlcptab = idxo->optionoutbwttab = NULL;
    idxo->sfxstrategy.spmopt_minlength = 0;
#ifndef S_SPLINT_S
    gt_registerPackedIndexOptions(op,
                                  &idxo->bwtIdxParams,
                                  BWTDEFOPT_CONSTRUCTION,
                                  idxo->indexname);
#endif
  }

  gt_option_parser_register_hook(op, gt_index_options_check_set_out_opts, idxo);

  return idxo;
}
Exemplo n.º 24
0
static GtOptionParser*
gt_condenseq_compress_option_parser_new(void *tool_arguments)
{
  GtCondenseqCompressArguments *arguments = tool_arguments;
  GtOptionParser *op;
  GtOption *option,
           *option_fraction;
  gt_assert(arguments);

  /* init */
  op = gt_option_parser_new("[options] INPUTENCSEQ",
                            "Compresses a GtEncseq to a UniqueEncseq.");

  /* -indexname */
  option = gt_option_new_string("indexname",
                                "path and basename of files to store",
                                arguments->indexname, NULL);
  gt_option_parser_add_option(op, option);

  /* -kmersize */
  option = gt_option_new_uint_min("kmersize",
                                  "kmer-size used for the seeds, default "
                                  "depends on alphabet size",
                                  &arguments->kmersize, GT_UNDEF_UINT, 2U);
  gt_option_parser_add_option(op, option);

  /* -windowsize */
  option = gt_option_new_uint("windowsize",
                              "Size of window in which to search for hit pairs "
                              "of kmers, has to be larger than kmersize" ,
                              &arguments->windowsize, GT_UNDEF_UINT);
  gt_option_parser_add_option(op, option);

  /* -initsize */
  option = gt_option_new_uword("initsize",
                               "length of inital unique database in bases, "
                               "should be larger than -alignlength",
                               &arguments->initsize, GT_UNDEF_UWORD);
  gt_option_parser_add_option(op, option);

  /* -alignlength */
  option = gt_option_new_uword("alignlength",
                               "required minimal length of an xdrop-alignment, "
                               "should be larger than -windowsize",
                               &arguments->minalignlength, GT_UNDEF_UWORD);
  gt_option_parser_add_option(op, option);

  /* -cutoff */
  option = gt_option_new_uword("cutoff",
                               "if a kmer is found more often than this value "
                               "it will be ignored for alignment searches. "
                               "Setting this to 0 will disable cutoffs, "
                               "leaving it undefined will use a cutoff based "
                               "on the mean number of occurences of a k-word.",
                               &arguments->cutoff_value, GT_UNDEF_UWORD);
  gt_option_is_extended_option(option);
  gt_option_parser_add_option(op, option);

  /* -fraction */
  option_fraction = gt_option_new_uword("fraction",
                               "when cutoffs aren'd disabled and no specific "
                               "value is set the mean number of occurrences "
                               "of each kmer divided by -fraction will be used "
                               "as cutoff",
                               &arguments->fraction, (GtUword) 2);
  gt_option_is_extended_option(option_fraction);
  gt_option_exclude(option, option_fraction);
  gt_option_parser_add_option(op, option_fraction);

  /* -disable_prune */
  option = gt_option_new_bool("disable_prune",
                              "when cutoffs and this option are set, "
                              "the database will still save every kmer, even "
                              "though only cutoff many kmers will be used.",
                              &arguments->prune, false);
  gt_option_is_extended_option(option);
  gt_option_parser_add_option(op, option);

  /* -mat */
  option = gt_option_new_int("mat",
                             "matchscore for extension-alignment, "
                             "requirements: mat > mis, mat > 2ins, mat > 2del",
                             &arguments->scores.mat, 2);
  gt_option_is_extended_option(option);
  gt_option_parser_add_option(op, option);

  /* -mis */
  option = gt_option_new_int("mis",
                             "mismatchscore for extension-alignment, ",
                             &arguments->scores.mis, -1);
  gt_option_is_extended_option(option);
  gt_option_parser_add_option(op, option);

  /* -ins */
  option = gt_option_new_int("ins",
                             "insertionscore for extension-alignment",
                             &arguments->scores.ins, -2);
  gt_option_is_extended_option(option);
  gt_option_parser_add_option(op, option);

  /* -del */
  option = gt_option_new_int("del",
                             "deletionscore for extension-alignment",
                             &arguments->scores.del, -2);
  gt_option_is_extended_option(option);
  gt_option_parser_add_option(op, option);

  /* -xdrop */
  option = gt_option_new_word("xdrop",
                              "xdrop score for extension-alignment",
                              &arguments->xdrop, (GtWord) 3);
  gt_option_is_extended_option(option);
  gt_option_parser_add_option(op, option);

  /* -brute_force */
  option = gt_option_new_bool("brute_force", "disable filtering of seeds. "
                              "Incompatible with -diagonals yes "
                              "or -full_diags yes",
                              &arguments->brute, false);
  gt_option_is_development_option(option);
  gt_option_parser_add_option(op, option);

  /* -diagonals */
  option = gt_option_new_bool("diagonals", "use sparse diagonals. "
                              "Incompatible with -brute_force yes. "
                              "Disabling both diagonals will result in simple "
                              "filtering of seed positions.",
                              &arguments->diags, false);
  gt_option_is_development_option(option);
  gt_option_parser_add_option(op, option);

  /* -full_diags */
  option = gt_option_new_bool("full_diags", "use full (time efficient "
                              "space inefficient) diagonals. "
                              "Incompatible with -brute_force yes. "
                              "Disabling both diagonals will result in simple "
                              "filtering of seed positions.",
                              &arguments->full_diags, false);
  gt_option_is_development_option(option);
  gt_option_parser_add_option(op, option);

  /* -clean_percent */
  option = gt_option_new_uint("diags_clean",
                              "Percentage of sparse diagonals that is allowed "
                              "to be marked as deletable. Sensible default is "
                              "set." ,
                              &arguments->clean_percent, GT_UNDEF_UINT);
  gt_option_is_development_option(option);
  gt_option_parser_add_option(op, option);

  /* -verbose */
  option = gt_option_new_bool("verbose", "enable verbose output",
                              &arguments->verbose, false);
  gt_option_parser_add_option(op, option);

  /* -kdb*/
  option = gt_option_new_bool("kdb", "prints out the kmer database (frequency "
                              "of each kmer), if -verbose each startposition "
                              "will be shown instead",
                              &arguments->kdb, false);
  gt_option_parser_add_option(op, option);

  return op;
}
Exemplo n.º 25
0
static GtIndexOptions* gt_index_options_register_generic_create(
                                                      GtOptionParser *op,
                                                      GtIndexOptionsIndexType t)
{
  GtIndexOptions *idxo;
  gt_assert(op != NULL && t != GT_INDEX_OPTIONS_UNDEFINED);
  idxo = gt_index_options_new();
  idxo->type = t;
  idxo->optionprefixlength = gt_option_new_uint_min("pl",
                                    "specify prefix length for bucket sort\n"
                                    "recommendation: use without argument;\n"
                                    "then a reasonable prefix length is "
                                    "automatically determined.",
                                    &idxo->prefixlength,
                                    GT_PREFIXLENGTH_AUTOMATIC,
                                    1U);
  gt_option_argument_is_optional(idxo->optionprefixlength);
  gt_option_parser_add_option(op, idxo->optionprefixlength);

  idxo->optionuserdefinedsortmaxdepth
    = gt_option_new_uint_min("sortmaxdepth","sort only up to the given depth.",
                             &idxo->sfxstrategy.userdefinedsortmaxdepth,
                             0,
                             1U);
  gt_option_parser_add_option(op, idxo->optionuserdefinedsortmaxdepth);
  gt_option_is_development_option(idxo->optionuserdefinedsortmaxdepth);

  idxo->optiondifferencecover = gt_option_new_uint_min("dc",
                                    "specify difference cover value",
                                    &idxo->sfxstrategy.differencecover,
                                    0,
                                    4U);
  gt_option_parser_add_option(op, idxo->optiondifferencecover);
  gt_option_exclude(idxo->optionuserdefinedsortmaxdepth,
                    idxo->optiondifferencecover);

  idxo->optioncmpcharbychar = gt_option_new_bool("cmpcharbychar",
                                           "compare suffixes character "
                                           "by character",
                                           &idxo->sfxstrategy.cmpcharbychar,
                                           false);
  gt_option_is_development_option(idxo->optioncmpcharbychar);
  gt_option_parser_add_option(op, idxo->optioncmpcharbychar);

  idxo->optionnoshortreadsort = gt_option_new_bool("noshortreadsort",
                                           "do not use short read sort",
                                           &idxo->sfxstrategy.noshortreadsort,
                                           false);
  gt_option_is_development_option(idxo->optionnoshortreadsort);
  gt_option_parser_add_option(op, idxo->optionnoshortreadsort);

  idxo->optionmaxwidthrealmedian = gt_option_new_uword("maxwidthrealmedian",
                                                 "compute real median for "
                                                 "intervals of at most the "
                                                 "given widthprefixes",
                                                 &idxo->sfxstrategy.
                                                      maxwidthrealmedian,
                                                 1UL);
  gt_option_is_development_option(idxo->optionmaxwidthrealmedian);
  gt_option_parser_add_option(op, idxo->optionmaxwidthrealmedian);

  idxo->optionalgbounds
    = gt_option_new_string_array("algbds",
                                 "length boundaries for the different "
                                 "algorithms to sort buckets of suffixes\n"
                                 "first number: maxbound for insertion sort\n"
                                 "second number: maxbound for blindtrie sort\n"
                                 "third number: maxbound for counting sort",
                                 idxo->algbounds);
  gt_option_is_development_option(idxo->optionalgbounds);
  gt_option_parser_add_option(op, idxo->optionalgbounds);

  idxo->optionstorespecialcodes
    = gt_option_new_bool("storespecialcodes",
                         "store special codes (this may speed up the program)",
                         &idxo->sfxstrategy.storespecialcodes,false);

  gt_option_is_development_option(idxo->optionstorespecialcodes);
  gt_option_parser_add_option(op, idxo->optionstorespecialcodes);

  idxo->optionparts
    = gt_option_new_uint_max("parts",
                             "specify number of parts in which the index "
                             "construction is performed",
                             &idxo->numofparts, 1U, (unsigned) ((1 << 22) - 1));
  gt_option_is_development_option(idxo->optionparts);
  gt_option_parser_add_option(op, idxo->optionparts);

  if (idxo->type == GT_INDEX_OPTIONS_ESA)
  {
    idxo->optionspmopt = gt_option_new_uint_min("spmopt",
                                            "optimize esa-construction for "
                                            "suffix-prefix matching",
                                            &idxo->sfxstrategy.spmopt_minlength,
                                            0,1U);
    gt_option_parser_add_option(op, idxo->optionspmopt);
    gt_option_exclude(idxo->optionspmopt, idxo->optiondifferencecover);
    idxo->optionmemlimit = gt_option_new_string("memlimit",
                           "specify maximal amount of memory to be used during "
                           "index construction (in bytes, the keywords 'MB' "
                           "and 'GB' are allowed)",
                           idxo->memlimit, NULL);
    gt_option_parser_add_option(op, idxo->optionmemlimit);
    gt_option_exclude(idxo->optionmemlimit, idxo->optionparts);
  }

  idxo->option = gt_option_new_bool("iterscan",
                                   "use iteratorbased-kmer scanning",
                                   &idxo->sfxstrategy.iteratorbasedkmerscanning,
                                   false);
  gt_option_is_development_option(idxo->option);
  gt_option_parser_add_option(op, idxo->option);

  idxo->option = gt_option_new_bool("samplewithprefixlengthnull",
                                  "sort sample with prefixlength=0",
                                  &idxo->sfxstrategy.samplewithprefixlengthnull,
                                  false);
  gt_option_is_development_option(idxo->option);
  gt_option_parser_add_option(op, idxo->option);

  idxo->option = gt_option_new_bool("suftabuint",
                                    "use uint32_t for suftab",
                                    &idxo->sfxstrategy.suftabuint,
                                    false);
  gt_option_is_development_option(idxo->option);
  gt_option_parser_add_option(op, idxo->option);

  idxo->option = gt_option_new_bool("onlybucketinsertion",
                                    "perform only bucket insertion",
                                    &idxo->sfxstrategy.onlybucketinsertion,
                                    false);
  gt_option_is_development_option(idxo->option);
  gt_option_parser_add_option(op, idxo->option);

  idxo->option = gt_option_new_bool("kmerswithencseqreader",
                               "always perform kmerscanning with encseq-reader",
                               &idxo->sfxstrategy.kmerswithencseqreader,
                               false);
  gt_option_is_development_option(idxo->option);
  gt_option_parser_add_option(op, idxo->option);

  idxo->option = gt_option_new_bool("dccheck",
                               "check intermediate results in difference cover",
                               &idxo->sfxstrategy.dccheck,
                               false);
  gt_option_is_development_option(idxo->option);
  gt_option_parser_add_option(op, idxo->option);

  idxo->option = gt_option_new_bool("withradixsort",
                                    "use radixsort to sort the buckets",
                                    &idxo->sfxstrategy.withradixsort,
                                    false);
  gt_option_is_development_option(idxo->option);
  gt_option_parser_add_option(op, idxo->option);

  idxo->option = gt_option_new_bool("compressedoutput",
                                    "output suftab and lcptab in compressed "
                                    "form",
                                    &idxo->sfxstrategy.compressedoutput,
                                    false);
  gt_option_is_development_option(idxo->option);
  gt_option_parser_add_option(op, idxo->option);

  gt_option_parser_register_hook(op, gt_index_options_check_set_create_opts,
                                 idxo);
  return idxo;
}
Exemplo n.º 26
0
static GtOptionParser* gtr_option_parser_new(GtR *gtr)
{
  GtOptionParser *op;
  GtOption *o, *only_option, *debug_option, *debugfp_option;
  gt_assert(gtr);
  op = gt_option_parser_new("[option ...] [tool | script] [argument ...]",
                            "The GenomeTools genome analysis system.");
  gt_option_parser_set_comment_func(op, show_gtr_help, gtr->tools);
  o = gt_option_new_bool("i",
                         "enter interactive mode after executing 'tool' or "
                         "'script'", &gtr->interactive, false);
  gt_option_hide_default(o);
  gt_option_parser_add_option(op, o);
  o = gt_option_new_bool("q", "suppress warnings", &gtr->quiet, false);
  gt_option_hide_default(o);
  gt_option_parser_add_option(op, o);
  o = gt_option_new_uint_min("j", "set number of parallel threads used at once",
                             &gt_jobs, 1, 1);
  gt_option_is_development_option(o);
  gt_option_parser_add_option(op, o);
  o = gt_option_new_bool("test", "perform unit tests and exit", &gtr->test,
                         false);
  gt_option_hide_default(o);
  gt_option_parser_add_option(op, o);
  only_option = gt_option_new_string("only", "perform single unit test "
                                     "(requires -test)", gtr->test_only, "");
  gt_option_imply(only_option, o);
  gt_option_is_development_option(only_option);
  gt_option_hide_default(only_option);
  gt_option_parser_add_option(op, only_option);
  debug_option = gt_option_new_debug(&gtr->debug);
  gt_option_parser_add_option(op, debug_option);
  debugfp_option = gt_option_new_string("debugfp",
                                     "set file pointer for debugging output\n"
                                     "use ``stdout'' for standard output\n"
                                     "use ``stderr'' for standard error\n"
                                     "or any other string to use the "
                                     "corresponding file (will be overwritten "
                                     "without warning!)", gtr->debugfp,
                                     "stderr");
  gt_option_is_development_option(debugfp_option);
  gt_option_parser_add_option(op, debugfp_option);
  gt_option_imply(debugfp_option, debug_option);
  o = gt_option_new_uint("seed",
                         "set seed for random number generator manually.\n"
                         "0 generates a seed from current time and process id",
                         &gtr->seed, gtr->seed);
  gt_option_hide_default(o);
  gt_option_parser_add_option(op, o);
  o = gt_option_new_bool("64bit", "exit with code 0 if this is a 64bit binary, "
                         "with 1 otherwise", &gtr->check64bit, false);
  gt_option_is_development_option(o);
  gt_option_parser_add_option(op, o);
  o = gt_option_new_bool("list", "list all tools and exit", &gtr->list, false);
  gt_option_is_development_option(o);
  gt_option_hide_default(o);
  gt_option_parser_add_option(op, o);
  o = gt_option_new_filename("testspacepeak", "alloc 64 MB and mmap the given "
                             "file", gtr->testspacepeak);
  gt_option_is_development_option(o);
  gt_option_parser_add_option(op, o);

  o = gt_option_new_string("createman", "create man page sources in directory",
                           gtr->manoutdir, "");
  gt_option_is_development_option(o);
  gt_option_parser_add_option(op, o);
  return op;
}
Exemplo n.º 27
0
static GtOptionParser* gt_gff3_option_parser_new(void *tool_arguments)
{
  GFF3Arguments *arguments = tool_arguments;
  GtOptionParser *op;
  GtOption *sort_option, *load_option, *strict_option, *tidy_option,
           *mergefeat_option, *addintrons_option, *offset_option,
           *offsetfile_option, *setsource_option, *option;
  gt_assert(arguments);

  /* init */
  op = gt_option_parser_new("[option ...] [GFF3_file ...]", "Parse, possibly "
                            "transform, and output GFF3 files.");

  /* -sort */
  sort_option = gt_option_new_bool("sort", "sort the GFF3 features (memory "
                                   "consumption is proportional to the input "
                                   "file size(s))",
                                   &arguments->sort, false);
  gt_option_parser_add_option(op, sort_option);

  /* -strict */
  strict_option = gt_option_new_bool("strict", "be very strict during GFF3 "
                                     "parsing (stricter than the specification "
                                     "requires)", &arguments->strict, false);
  gt_option_is_development_option(strict_option);
  gt_option_parser_add_option(op, strict_option);

  /* -tidy */
  tidy_option = gt_option_new_bool("tidy", "try to tidy the GFF3 files up "
                                   "during parsing", &arguments->tidy, false);
  gt_option_parser_add_option(op, tidy_option);
  gt_option_exclude(strict_option, tidy_option);

  /* -retainids */
  option = gt_option_new_bool("retainids",
                              "when available, use the original IDs provided "
                              "in the source file\n"
                              "(memory consumption is proportional to the "
                              "input file size(s))", &arguments->retainids,
                              false);
  gt_option_parser_add_option(op, option);

  /* -checkids */
  option = gt_option_new_bool("checkids",
                              "make sure the ID attributes are unique "
                              "within the scope of each GFF3_file, as required "
                              "by GFF3 specification\n"
                              "(memory consumption is proportional to the "
                              "input file size(s))", &arguments->checkids,
                              false);
  gt_option_parser_add_option(op, option);

  /* -addids */
  option = gt_option_new_bool("addids", "add missing \""
                              GT_GFF_SEQUENCE_REGION"\" lines automatically",
                              &arguments->addids, true);
  gt_option_parser_add_option(op, option);

  /* -fixregionboundaries */
  option = gt_option_new_bool("fixregionboundaries", "automatically adjust \""
                              GT_GFF_SEQUENCE_REGION"\" lines to contain all "
                              "their features (memory consumption is "
                              "proportional to the input file size(s))",
                              &arguments->fixboundaries, false);
  gt_option_parser_add_option(op, option);

  /* -mergefeat */
  mergefeat_option = gt_option_new_bool("mergefeat",
                                        "merge adjacent features of the same "
                                        "type", &arguments->mergefeat, false);
  gt_option_is_development_option(mergefeat_option);
  gt_option_imply(mergefeat_option, sort_option);
  gt_option_parser_add_option(op, mergefeat_option);

  /* -load */
  load_option = gt_option_new_bool("load", "load the GFF3 features into memory "
                                   "(requires space proportional to the input "
                                   "file size(s))",
                                   &arguments->load, false);
  gt_option_is_development_option(load_option);
  gt_option_parser_add_option(op, load_option);

  /* -addintrons */
  addintrons_option = gt_option_new_bool("addintrons", "add intron features "
                                         "between existing exon features",
                                         &arguments->addintrons, false);
  gt_option_parser_add_option(op, addintrons_option);

  /* -offset */
  offset_option = gt_option_new_word("offset", "transform all features by the "
                                     "given offset", &arguments->offset,
                                     GT_UNDEF_WORD);
  gt_option_parser_add_option(op, offset_option);

  /* -offsetfile */
  offsetfile_option = gt_option_new_filename("offsetfile", "transform all "
                                             "features by the offsets given in "
                                             "file", arguments->offsetfile);
  gt_option_parser_add_option(op, offsetfile_option);
  gt_option_exclude(offset_option, offsetfile_option);

  /* -setsource */
  setsource_option = gt_option_new_string("setsource", "set the 'source' "
                                          "value (2nd column) of each feature",
                                          arguments->newsource, NULL);
  gt_option_parser_add_option(op, setsource_option);

  /* typecheck options */
  gt_typecheck_info_register_options(arguments->tci, op);

  /* -show */
  option = gt_option_new_bool("show", "show GFF3 output", &arguments->show,
                              true);
  gt_option_parser_add_option(op, option);

  /* -v */
  option = gt_option_new_verbose(&arguments->verbose);
  gt_option_parser_add_option(op, option);

  /* -width */
  option = gt_option_new_width(&arguments->width);
  gt_option_parser_add_option(op, option);

  /* output file options */
  gt_output_file_info_register_options(arguments->ofi, op, &arguments->outfp);

  /* set comment function */
  gt_option_parser_set_comment_func(op, gt_gtdata_show_help, NULL);

  return op;
}
Exemplo n.º 28
0
static GtOptionParser* gt_gdiffcalc_option_parser_new(void *tool_arguments)
{
  GtGenomediffArguments *arguments = tool_arguments;
  GtOptionParser *op;
  GtOption *option, *option_unitfile;

  gt_assert(arguments);

  /* init */
  op = gt_option_parser_new("[option ...] "
                          "-indexname NAME AVGSHULEN) ",
                          "Calculates Kr: pairwise distances between genomes.");

  /* options */
  option = gt_option_new_string("indexname", "Basename of encseq to construct.",
                                arguments->indexname, NULL);
  gt_option_is_mandatory(option);
  gt_option_parser_add_option(op, option);

  /*-unitfile*/
  option_unitfile =
    gt_option_new_filename("unitfile",
                           "specifies genomic units, see below for description",
                           arguments->unitfile);
  gt_option_parser_add_option(op, option_unitfile);
  arguments->ref_unitfile = gt_option_ref(option_unitfile);

  /* encseq options */
  arguments->loadopts =
    gt_encseq_options_register_loading(op, arguments->indexname);

  gt_option_is_development_option(
                        gt_encseq_options_lossless_option(arguments->loadopts));

  /* dev options */
  /* -max_n */
  option = gt_option_new_uword("max_n", "Number of precalculated values "
                               "for ln(n!) and pmax(x)",
                               &arguments->max_ln_n_fac, 1000UL);
  gt_option_is_development_option(option);
  gt_option_parser_add_option(op, option);

  /* thresholds */
  /* divergence error */
  option = gt_option_new_double("thr",
                                "Threshold for difference (du, dl) in "
                                "divergence calculation.\n"
                                "default: 1e-9",
                                &arguments->divergence_threshold,
                                1e-9);
  gt_option_is_extended_option(option);
  gt_option_hide_default(option);
  gt_option_parser_add_option(op, option);

  /* expected shulen error */
  option = gt_option_new_double("abs_err",
                                "absolute error for expected shulen "
                                "calculation.\n"
                                "default: 1e-5",
                                &arguments->divergence_abs_err,
                                1e-5);
  gt_option_is_extended_option(option);
  gt_option_hide_default(option);
  gt_option_parser_add_option(op, option);

  /* relative expected shulen error */
  option = gt_option_new_double("rel_err",
                                "relative error for expected shulen "
                                "calculation.\n"
                                "default: 1e-3",
                                &arguments->divergence_rel_err,
                                1e-3);
  gt_option_is_extended_option(option);
  gt_option_hide_default(option);
  gt_option_parser_add_option(op, option);

  /* M */
  option = gt_option_new_double("M",
                                "threshold for minimum logarithm.\n"
                                "default: DBL_MIN",
                                &arguments->divergence_m,
                                DBL_MIN);
  gt_option_is_extended_option(option);
  gt_option_hide_default(option);
  gt_option_parser_add_option(op, option);

  /* -v */
  option = gt_option_new_verbose(&arguments->verbose);
  gt_option_parser_add_option(op, option);

  /* mail */
  gt_option_parser_set_mail_address(op, "<*****@*****.**>");
  /* doc */
  gt_option_parser_set_comment_func(op, gt_gtdata_show_help, NULL);
  return op;
}
Exemplo n.º 29
0
static GtOptionParser* gt_seed_extend_option_parser_new(void *tool_arguments)
{
  GtSeedExtendArguments *arguments = tool_arguments;
  GtOptionParser *op;
  GtOption *option, *op_gre, *op_xdr, *op_cam, *op_his, *op_dif, *op_pmh,
    *op_len, *op_err, *op_xbe, *op_sup, *op_frq, *op_mem, *op_ali, *op_bia;
  gt_assert(arguments != NULL);

  /* init */
  op = gt_option_parser_new("[option ...] encseq_basename [encseq_basename]",
                            "Calculate local alignments using the seed and "
                            "extend algorithm.");

  /* DIAGBANDSEED OPTIONS */

  /* -ii */
  option = gt_option_new_string("ii",
                                "Input index for encseq encoded sequences",
                                arguments->dbs_indexname,
                                "");
  gt_option_is_mandatory(option);
  gt_option_parser_add_option(op, option);

  /* -qii */
  option = gt_option_new_string("qii",
                                "Query input index (encseq)",
                                arguments->dbs_queryname,
                                "");
  gt_option_parser_add_option(op, option);

  /* -seedlength */
  op_len = gt_option_new_uint_min_max("seedlength",
                                      "Minimum length of a seed",
                                      &arguments->dbs_seedlength,
                                      14UL, 1UL, 32UL);
  gt_option_parser_add_option(op, op_len);

  /* -diagbandwidth */
  option = gt_option_new_uword("diagbandwidth",
                               "Logarithm of diagonal band width (for filter)",
                               &arguments->dbs_logdiagbandwidth,
                               6UL);
  gt_option_parser_add_option(op, option);

  /* -mincoverage */
  option = gt_option_new_uword("mincoverage",
                               "Minimum coverage in two neighbouring diagonal "
                               "bands (for filter)",
                               &arguments->dbs_mincoverage,
                               35UL);
  gt_option_parser_add_option(op, option);

  /* -maxfreq */
  op_frq = gt_option_new_uword_min("maxfreq",
                                   "Maximum frequency of a k-mer (for filter)",
                                   &arguments->dbs_maxfreq,
                                   GT_UWORD_MAX, 1UL);
  gt_option_parser_add_option(op, op_frq);

  /* -t */
  op_sup = gt_option_new_uword_min("t",
                                   "Suppress k-mers occurring at least t times "
                                   "(for filter)",
                                   &arguments->dbs_suppress,
                                   GT_UWORD_MAX, 2UL);
  gt_option_exclude(op_sup, op_frq);
  gt_option_is_development_option(op_sup);
  gt_option_parser_add_option(op, op_sup);

  /* -memlimit */
  op_mem = gt_option_new_string("memlimit",
                                "Maximum memory usage to determine the maximum "
                                "frequency of a k-mer (for filter)",
                                arguments->dbs_memlimit_str,
                                "");
  gt_option_parser_add_option(op, op_mem);

  /* -debug-kmer */
  option = gt_option_new_bool("debug-kmer",
                              "Output KmerPos lists",
                              &arguments->dbs_debug_kmer,
                              false);
  gt_option_is_development_option(option);
  gt_option_parser_add_option(op, option);

  /* -debug-seedpair */
  option = gt_option_new_bool("debug-seedpair",
                              "Output SeedPair lists",
                              &arguments->dbs_debug_seedpair,
                              false);
  gt_option_is_development_option(option);
  gt_option_parser_add_option(op, option);

  /* -verify */
  option = gt_option_new_bool("verify",
                              "Check that k-mer seeds occur in the sequences",
                              &arguments->dbs_verify,
                              false);
  gt_option_is_development_option(option);
  gt_option_parser_add_option(op, option);

  /* SEED EXTENSION OPTIONS */

  /* -extendxdrop */
  op_xdr = gt_option_new_uword_min_max("extendxdrop",
                                       "Extend seed to both sides using xdrop "
                                       "algorithm, optional parameter "
                                       "specifies sensitivity",
                                       &arguments->se_extendxdrop,
                                       97UL, 90UL, 100UL);
  gt_option_argument_is_optional(op_xdr);
  gt_option_parser_add_option(op, op_xdr);
  arguments->se_option_xdrop = gt_option_ref(op_xdr);

  /* -xdropbelow */
  op_xbe = gt_option_new_word("xdropbelow",
                              "Specify xdrop cutoff score (0 means "
                              "automatically defined depending on minidentity)",
                              &arguments->se_xdropbelowscore,
                              0L);
  gt_option_imply(op_xbe, op_xdr);
  gt_option_parser_add_option(op, op_xbe);

  /* -extendgreedy */
  op_gre = gt_option_new_uword_min_max("extendgreedy",
                                       "Extend seed to both sides using greedy "
                                       "algorithm, optional parameter "
                                       "specifies sensitivity",
                                       &arguments->se_extendgreedy,
                                       97UL, 90UL, 100UL);
  gt_option_argument_is_optional(op_gre);
  gt_option_exclude(op_gre, op_xdr);
  gt_option_parser_add_option(op, op_gre);
  arguments->se_option_greedy = gt_option_ref(op_gre);

  /* -history */
  op_his = gt_option_new_uword_min_max("history",
                                       "Size of (mis)match history in range [1"
                                       "..64] (trimming for greedy extension)",
                                       &arguments->se_historysize,
                                       60UL, 1UL, 64UL);
  gt_option_imply(op_his, op_gre);
  gt_option_parser_add_option(op, op_his);

  /* -maxalilendiff */
  op_dif = gt_option_new_uword("maxalilendiff",
                               "Maximum difference of alignment length "
                               "(trimming for greedy extension)",
                               &arguments->se_maxalilendiff, 0UL);
  gt_option_imply(op_dif, op_gre);
  gt_option_is_development_option(op_dif);
  gt_option_parser_add_option(op, op_dif);

  /* -percmathistory */
  op_pmh = gt_option_new_uword_min_max("percmathistory",
                                       "percentage of matches required in "
                                       "history (for greedy extension)",
                                       &arguments->se_perc_match_hist,
                                       0UL, 1UL, 100UL);
  gt_option_imply(op_pmh, op_gre);
  gt_option_is_development_option(op_pmh);
  gt_option_parser_add_option(op, op_pmh);

  /* -bias-parameters */
  op_bia = gt_option_new_bool("bias-parameters",
                              "Use -maxalilendiff 30 and let percmathistory "
                              "depend on minidentiy and DNA base distribution",
                              &arguments->bias_parameters,
                              false);
  gt_option_imply(op_bia, op_gre);
  gt_option_exclude(op_bia, op_pmh);
  gt_option_exclude(op_bia, op_dif);
  gt_option_is_development_option(op_bia);
  gt_option_parser_add_option(op, op_bia);

  /* -cam */
  op_cam = gt_option_new_string("cam",
                                gt_cam_extendgreedy_comment(),
                                arguments->se_char_access_mode,
                                "");
  gt_option_is_development_option(op_cam);
  gt_option_parser_add_option(op, op_cam);

  /* -l */
  op_len = gt_option_new_uword_min("l",
                                   "Minimum alignment length "
                                   "(for seed extension)",
                                   &arguments->se_alignlength,
                                   20UL, 1UL);
  gt_option_imply_either_2(op_len, op_xdr, op_gre);
  gt_option_parser_add_option(op, op_len);

  /* -minidentity */
  op_err = gt_option_new_uword_min_max("minidentity",
                                       "Minimum identity of matches "
                                       "(for seed extension)",
                                       &arguments->se_minidentity,
                                       80UL, GT_EXTEND_MIN_IDENTITY_PERCENTAGE,
                                       99UL);
  gt_option_imply_either_2(op_err, op_xdr, op_gre);
  gt_option_parser_add_option(op, op_err);

  /* -a */
  op_ali = gt_option_new_uword_min("a",
                                   "show alignments/sequences (optional "
                                   "argument is number of columns per line)",
                                   &arguments->se_alignmentwidth,
                                   70, 20);
  gt_option_argument_is_optional(op_ali);
  gt_option_parser_add_option(op, op_ali);
  arguments->se_option_withali = gt_option_ref(op_ali);

  /* -mirror */
  option = gt_option_new_bool("mirror",
                              "Add reverse complement reads",
                              &arguments->mirror,
                              false);
  gt_option_parser_add_option(op, option);

  /* -overlappingseeds */
  option = gt_option_new_bool("overlappingseeds",
                              "Allow overlapping SeedPairs",
                              &arguments->overlappingseeds,
                              false);
  gt_option_is_development_option(option);
  gt_option_parser_add_option(op, option);

  /* -benchmark */
  option = gt_option_new_bool("benchmark",
                              "Measure total running time and be silent",
                              &arguments->benchmark,
                              false);
  gt_option_is_development_option(option);
  gt_option_parser_add_option(op, option);

  /* -seed-display */
  option = gt_option_new_bool("seed-display",
                              "Display seeds in #-line",
                              &arguments->seed_display,
                              false);
  gt_option_is_development_option(option);
  gt_option_parser_add_option(op, option);

  /* -v */
  option = gt_option_new_verbose(&arguments->verbose);
  gt_option_parser_add_option(op, option);

  return op;
}
Exemplo n.º 30
0
static GtOptionParser* gt_eval_option_parser_new(void *tool_arguments)
{
  EvalArguments *arguments = tool_arguments;
  GtOptionParser *op;
  GtOption *option, *ltroption, *ltrdeltaoption;
  gt_assert(arguments);

  op = gt_option_parser_new("reference_file prediction_file ",
                            "Compare annotation files and show "
                            "accuracy measures (prediction vs. reference).");

  /* -exondiff */
  option = gt_option_new_bool("exondiff", "show a diff for the exons",
                              &arguments->exondiff, false);
  gt_option_is_development_option(option);
  gt_option_parser_add_option(op, option);

  /* -exondiffcollapsed */
  option = gt_option_new_bool("exondiffcollapsed", "show a diff for the "
                              "collapsed exons", &arguments->exondiffcollapsed,
                              false);
  gt_option_is_development_option(option);
  gt_option_parser_add_option(op, option);

  /* -nuc */
  option = gt_option_new_bool("nuc",
                              "evaluate nucleotide level (memory consumption "
                              "is proportional to the input file sizes)",
                              &arguments->nuceval, true);
  gt_option_parser_add_option(op, option);

  /* -ltr */
  ltroption = gt_option_new_bool("ltr", "evaluate a LTR retrotransposon "
                                 "prediction instead of a gene prediction\n"
                                 "(all LTR_retrotransposon elements are "
                                 "considered to have an undetermined strand)",
                                 &arguments->evalLTR, false);
  gt_option_parser_add_option(op, ltroption);

  /* -ltrdelta */
  ltrdeltaoption = gt_option_new_uword("ltrdelta", "set allowed delta for LTR "
                                       "borders to be considered equal",
                                       &arguments->LTRdelta, 20);
  gt_option_parser_add_option(op, ltrdeltaoption);

  /* -v */
  option = gt_option_new_verbose(&arguments->verbose);
  gt_option_parser_add_option(op, option);

  /* output file options */
  gt_output_file_info_register_options(arguments->ofi, op, &arguments->outfp);

  /* option implications */
  gt_option_imply(ltrdeltaoption, ltroption);

  /* set comment function */
  gt_option_parser_set_comment_func(op, gt_gtdata_show_help, NULL);

  /* set minimum and maximum number of arguments */
  gt_option_parser_set_min_max_args(op, 2, 2);

  return op;
}