static GtOptionParser* gt_seqfilter_option_parser_new(void *tool_arguments)
{
SeqFilterArguments *arguments = tool_arguments;
GtOption *option;
GtOptionParser *op;
gt_assert(arguments);
op = gt_option_parser_new("[option ...] [sequence_file ...]",
"Filter the given sequence file(s) and show the "
"results on stdout.");
/* -minlength */
option = gt_option_new_uword("minlength",
"set minimum length a sequence must "
"have to pass the filter", &arguments->minlength,
GT_UNDEF_UWORD);
gt_option_parser_add_option(op, option);
/* -maxlength */
option = gt_option_new_uword("maxlength", "set maximum length a sequence can "
"have to pass the filter", &arguments->maxlength,
GT_UNDEF_UWORD);
gt_option_parser_add_option(op, option);
/* -maxseqnum */
option = gt_option_new_uword("maxseqnum", "set the maximum number of "
"sequences which can pass the filter",
&arguments->maxseqnum, GT_UNDEF_UWORD);
gt_option_parser_add_option(op, option);
/* -sample */
option = gt_option_new_double_min_max("sample", "set a probability for each "
"sequence to pass the filter",
&arguments->sample_prob,
SEQFILTER_DEF_PROB,
SEQFILTER_MIN_PROB,
SEQFILTER_MAX_PROB);
gt_option_parser_add_option(op, option);
/* -step */
option = gt_option_new_uword_min("step", "only every 'step'-th sequence "
"passes the filter",
&arguments->step,
SEQFILTER_DEF_STEP,
SEQFILTER_MIN_STEP);
gt_option_parser_add_option(op, option);
/* -width */
option = gt_option_new_width(&arguments->width);
gt_option_parser_add_option(op, option);
gt_output_file_info_register_options(arguments->ofi, op, &arguments->outfp);
return op;
}
static GtOptionParser* gt_splitfasta_option_parser_new(void *tool_arguments)
{
SplitfastaArguments *arguments = tool_arguments;
GtOptionParser *op;
GtOption *targetsize_option, *splitdesc_option, *o, *numfiles_option;
gt_assert(arguments);
op = gt_option_parser_new("[option ...] fastafile","Split the supplied fasta "
"file.");
numfiles_option = gt_option_new_uint_min("numfiles",
"set the number of target files "
"",
&arguments->num_files, 0,
1);
gt_option_parser_add_option(op, numfiles_option);
targetsize_option = gt_option_new_uword_min("targetsize",
"set the target file "
"size in MB",
&arguments->max_filesize_in_MB,
50, 1);
gt_option_parser_add_option(op, targetsize_option);
splitdesc_option = gt_option_new_string("splitdesc",
"put every fasta entry in "
"a separate file named by its "
"description in the given directory",
arguments->splitdesc, NULL);
gt_option_parser_add_option(op, splitdesc_option);
gt_option_exclude(targetsize_option, splitdesc_option);
gt_option_exclude(numfiles_option, splitdesc_option);
gt_option_exclude(numfiles_option, targetsize_option);
o = gt_option_new_width(&arguments->width);
gt_option_parser_add_option(op, o);
o = gt_option_new_bool(GT_FORCE_OPT_CSTR, "force writing to output file",
&arguments->force, false);
gt_option_parser_add_option(op, o);
gt_option_parser_set_min_max_args(op, 1, 1);
return op;
}
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;
}
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;
}
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;
}
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;
}
static GtOptionParser* gt_kmer_database_option_parser_new(void *tool_arguments)
{
GtKmerDatabaseArguments *arguments = tool_arguments;
GtOptionParser *op;
GtOption *option,
*option_verbose,
*option_use_cutoff,
*option_hash,
*option_mean_cutoff;
gt_assert(arguments);
/* init */
op = gt_option_parser_new("[option ...] [file]",
"Makes a GtKmerDatabase from the input file.");
/* -kmersize */
option = gt_option_new_uint_min_max("kmersize", "kmersize used",
&arguments->kmersize, 3U, 1U, 10U);
gt_option_parser_add_option(op, option);
/* -verbose */
option_verbose = gt_option_new_bool("verbose", "prints out results of "
"merging",
&arguments->verbose, false);
gt_option_parser_add_option(op, option_verbose);
/* -merge_only */
option = gt_option_new_bool("merge_only", "only uses merge to build DB, "
"doesn_t build two DBs to compare merge with a "
"different method (much faster). It also allows "
"for random intervals which are biffer than the "
"maximum buffer size (will be split internally).",
&arguments->merge_only, false);
gt_option_parser_add_option(op, option);
/* -use_cutoff */
option_use_cutoff = gt_option_new_bool("use_cutoff", "uses a cutoff. see "
"-set_cutoff description. Only works "
"with merge_only",
&arguments->cutoff, false);
gt_option_parser_add_option(op, option_use_cutoff);
gt_option_imply(option_use_cutoff, option);
/* -set_cutoff */
option = gt_option_new_uword_min("set_cutoff", "kmers occuring more often "
"than this value won't be saved",
&arguments->cutoff_value, (GtUword) 30,
(GtUword) 1);
gt_option_parser_add_option(op, option);
gt_option_imply(option, option_use_cutoff);
/* -mean_cutoff */
option_mean_cutoff = gt_option_new_bool("mean_cutoff", "2*mean of kmer "
"occurrence will be"
" used as cutoff value",
&arguments->mean_cutoff, false);
gt_option_parser_add_option(op, option_mean_cutoff);
gt_option_imply(option_mean_cutoff, option_use_cutoff);
gt_option_exclude(option_mean_cutoff, option);
/* -disable_prune */
option = gt_option_new_bool("disable_prune", "disables the removel of kmers, "
"which occure more often than the cutoff.",
&arguments->prune, false);
gt_option_parser_add_option(op, option);
gt_option_imply(option, option_use_cutoff);
/* -use_hash */
option_hash = gt_option_new_bool("use_hash", "saves each kmer in kdb and "
"also in a hash. afterwards both will be "
"accessed and time for saving and "
"accessing will be shown.",
&arguments->use_hash, false);
gt_option_parser_add_option(op, option_hash);
gt_option_exclude(option_hash, option_use_cutoff);
gt_option_exclude(option_hash, option_verbose);
/* -benchmark */
option = gt_option_new_bool("benchmark", "measures the time the tool takes to"
" fill the database. Doesn't test for consistency"
" though!", &arguments->bench, false);
gt_option_parser_add_option(op, option);
/* -bsize */
option = gt_option_new_uword_min("bsize", "size of the buffer",
&arguments->sb_size, (GtUword) 100000,
(GtUword) 2);
gt_option_parser_add_option(op, option);
gt_option_exclude(option, option_hash);
/* -outfile */
option = gt_option_new_string("outfile", "specifies file for verbose output",
arguments->print_filename, NULL);
gt_option_parser_add_option(op, option);
return op;
}
static GtOptionParser* gt_tir_option_parser_new(void *tool_arguments)
{
GtTirArguments *arguments = tool_arguments;
GtOptionParser *op;
GtOption *optionindex, /* index */
*optionseed, /* minseedlength */
*optionminlentir, /* minimal length of TIR */
*optionmaxlentir, /* maximal length of TIR */
*optionmindisttir, /* minimal distance of TIRs */
*optionmaxdisttir, /* maximal distance of TIRs */
*optionmat, /* arbitrary scores */
*optionmis,
*optionins,
*optiondel,
*optionxdrop, /* xdropbelowscore for extension alignment */
*optionsimilar, /* similarity threshold */
*optionoverlaps, /* for overlaps */
*optionmintsd, /* minimal length for Target Site Duplication */
*optionmaxtsd, /* maximal length for Target Site Duplication */
*optionvicinity, /* vicinity around TIRs to be searched for TSDs */
*optionhmms,
*optionevalcutoff,
*optionpdomcutoff,
*optionmaxgap;
static const char *overlaps[] = {
"best", /* default */
"no",
"all",
NULL
};
static const char *cutoffs[] = {
"NONE",
"GA",
"TC",
NULL
};
gt_assert(arguments);
/* init */
op = gt_option_parser_new("[option ...] -index INDEXNAME",
"Identify Terminal Inverted Repeat (TIR) elements,"
"such as DNA transposons.");
/* -index */
optionindex = gt_option_new_string("index",
"specify the name of the enhanced suffix "
"array index (mandatory)",
arguments->str_indexname, NULL);
gt_option_is_mandatory(optionindex);
gt_option_parser_add_option(op, optionindex);
/* -seed */
optionseed = gt_option_new_uword_min("seed",
"specify minimum seed length for "
"exact repeats",
&arguments->min_seed_length, 20UL, 2UL);
gt_option_parser_add_option(op, optionseed);
/* -minlentir */
optionminlentir = gt_option_new_uword_min_max("mintirlen",
"specify minimum length for "
"each TIR",
&arguments->min_TIR_length,
27UL, 1UL, GT_UNDEF_UWORD);
gt_option_parser_add_option(op, optionminlentir);
/* -maxlentir */
optionmaxlentir = gt_option_new_uword_min_max("maxtirlen",
"specify maximum length for "
"each TIR",
&arguments->max_TIR_length,
1000UL, 1UL, GT_UNDEF_UWORD);
gt_option_parser_add_option(op, optionmaxlentir);
/* -mindisttir */
optionmindisttir = gt_option_new_uword_min_max("mintirdist",
"specify minimum distance of "
"TIRs",
&arguments->min_TIR_distance,
100UL, 1UL, GT_UNDEF_UWORD);
gt_option_parser_add_option(op, optionmindisttir);
/* -maxdisttir */
optionmaxdisttir = gt_option_new_uword_min_max("maxtirdist",
"specify maximum distance of "
"TIRs",
&arguments->max_TIR_distance,
10000UL, 1UL, GT_UNDEF_UWORD);
gt_option_parser_add_option(op, optionmaxdisttir);
optionmat = gt_option_new_int_min("mat",
"specify matchscore for "
"extension-alignment",
&arguments->arbit_scores.mat, 2, 1);
gt_option_parser_add_option(op, optionmat);
/* -mis */
optionmis = gt_option_new_int_max("mis",
"specify mismatchscore for "
"extension-alignment",
&arguments->arbit_scores.mis, -2, -1);
gt_option_parser_add_option(op, optionmis);
/* -ins */
optionins = gt_option_new_int_max("ins",
"specify insertionscore for "
"extension-alignment",
&arguments->arbit_scores.ins, -3, -1);
gt_option_parser_add_option(op, optionins);
/* -del */
optiondel = gt_option_new_int_max("del",
"specify deletionscore for "
"extension-alignment",
&arguments->arbit_scores.del, -3, -1);
gt_option_parser_add_option(op, optiondel);
/* -xdrop */
optionxdrop = gt_option_new_int_min("xdrop",
"specify xdropbelowscore for "
"extension-alignment",
&arguments->xdrop_belowscore, (int) 5,
(int) 0);
gt_option_parser_add_option(op, optionxdrop);
/* -similar */
optionsimilar = gt_option_new_double_min_max("similar",
"specify similaritythreshold in "
"range [1..100%]",
&arguments->similarity_threshold,
(double) 85.0, (double) 0.0,
100.0);
gt_option_parser_add_option(op, optionsimilar);
/* -overlaps */
optionoverlaps = gt_option_new_choice("overlaps", "specify no|best|all",
arguments->str_overlaps,
overlaps[0], overlaps);
gt_option_parser_add_option(op, optionoverlaps);
arguments->optionoverlaps = gt_option_ref(optionoverlaps);
/* -mintsd */
optionmintsd = gt_option_new_uword_min_max("mintsd",
"specify minimum length for each "
"TSD",
&arguments->min_TSD_length,
2U, 0, GT_UNDEF_UINT);
gt_option_parser_add_option(op, optionmintsd);
/* -maxtsd */
optionmaxtsd = gt_option_new_uword_min_max("maxtsd",
"specify maximum length for each "
"TSD",
&arguments->max_TSD_length,
11U, 0, GT_UNDEF_UINT);
gt_option_parser_add_option(op, optionmaxtsd);
gt_option_imply(optionmaxtsd, optionmintsd);
/* -vicinity */
optionvicinity = gt_option_new_uword_min_max("vic",
"specify the number of "
"nucleotides (to the left and "
"to the right) that will be "
"searched for TSDs around 5' "
"and 3' boundary of predicted "
"TIRs",
&arguments->vicinity,
60U, 1U, 500U);
gt_option_parser_add_option(op, optionvicinity);
optionhmms = 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, optionhmms);
optionevalcutoff = 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, optionevalcutoff);
gt_option_hide_default(optionevalcutoff);
gt_option_imply(optionevalcutoff, optionhmms);
optionpdomcutoff = gt_option_new_choice("pdomcutoff",
"model-specific score cutoff\n"
"choose from TC (trusted cutoff) | "
"GA (gathering cutoff) | "
"NONE (no cutoffs)",
arguments->cutoffs, cutoffs[1], cutoffs);
gt_option_parser_add_option(op, optionpdomcutoff);
gt_option_imply(optionpdomcutoff, optionhmms);
/* XXX: make -pdomcutoff and -pdomevalcutoff mutually exclusive */
optionmaxgap = 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, optionmaxgap);
gt_option_is_extended_option(optionmaxgap);
gt_option_imply(optionmaxgap, optionhmms);
return op;
}
static GtOptionParser* gt_sketch_page_option_parser_new(void *tool_arguments)
{
SketchPageArguments *arguments = tool_arguments;
GtOptionParser *op;
static const char *formats[] = {
#ifdef CAIRO_HAS_PDF_SURFACE
"pdf",
#endif
#ifdef CAIRO_HAS_PS_SURFACE
"ps",
#endif
NULL
};
GtOption *o;
op = gt_option_parser_new("outfile annotationfile",
"Draw a multi-page PDF/PS representation of "
"an annotation file.");
o = gt_option_new_string("seqid", "sequence region to draw\n"
"default: first in file",
arguments->seqid, NULL);
gt_option_parser_add_option(op, o);
gt_option_hide_default(o);
o = gt_option_new_string("text", "text to show in header\n"
"default: file name",
arguments->text, NULL);
gt_option_parser_add_option(op, o);
gt_option_hide_default(o);
o = gt_option_new_double("fontsize", "header and footer font size "
"(in points)",
&arguments->theight, 10.0);
gt_option_parser_add_option(op, o);
o = gt_option_new_range("range", "range to draw (e.g. 100 10000)\n"
"default: full range",
&arguments->range, NULL);
gt_option_parser_add_option(op, o);
gt_option_hide_default(o);
o = gt_option_new_uword_min("linewidth", "base width of a single "
"repeated unit",
&arguments->width, 2000, 1000);
gt_option_is_mandatory(o);
gt_option_parser_add_option(op, o);
o = gt_option_new_double("width", "page width in millimeters "
"(default: DIN A4)",
&arguments->pwidth, 210.0);
gt_option_parser_add_option(op, o);
o = gt_option_new_double("height", "page height in millimeters "
"(default: DIN A4)",
&arguments->pheight, 297.0);
gt_option_parser_add_option(op, o);
o = gt_option_new_choice("format", "output format\n"
"choose from: "
#ifdef CAIRO_HAS_PDF_SURFACE
"pdf"
#ifdef CAIRO_HAS_PS_SURFACE
"|"
#endif
#endif
#ifdef CAIRO_HAS_PS_SURFACE
"ps"
#endif
"",
arguments->format, formats[0], formats );
gt_option_parser_add_option(op, o);
o = gt_option_new_string("style", "style file to use\n"
"default: gtdata/sketch/default.style",
arguments->stylefile,
gt_str_get(arguments->stylefile));
gt_option_parser_add_option(op, o);
gt_option_hide_default(o);
o = gt_option_new_filename("seqfile", "sequence file for GC content view",
arguments->seqfile);
gt_option_parser_add_option(op, o);
gt_option_is_extended_option(o);
gt_option_parser_set_min_max_args(op, 2, 2);
return op;
}
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;
}
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_mail_address(op, "<*****@*****.**>");
option = gt_option_new_filename_array("q","Specify files containing the "
"query sequences",
arguments->queryfiles);
gt_option_parser_add_option (op, option);
option = gt_option_new_word("match",
"Specify match score",
&arguments->matchscore, 1L);
gt_option_parser_add_option(op, option);
option = gt_option_new_word("mismatch",
"Specify mismatch score",
&arguments->mismatchscore, -3L);
gt_option_parser_add_option(op, option);
option = gt_option_new_word("gapstart",
"Specify gap start score",
&arguments->gapstart, -5L);
gt_option_parser_add_option(op, option);
option = gt_option_new_word("gapextend",
"Specify gap extension score",
&arguments->gapextend, -2L);
gt_option_parser_add_option(op, option);
option = gt_option_new_uword_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;
}
