void gt_output_file_info_register_options(GtOutputFileInfo *ofi,
GtOptionParser *op, GtFile **outfp)
{
GtOption *opto, *optgzip, *optbzip2, *optforce;
gt_assert(outfp && ofi);
ofi->outfp = outfp;
/* register option -o */
opto = gt_option_new_filename("o", "redirect output to specified file",
ofi->output_filename);
gt_option_parser_add_option(op, opto);
/* register option -gzip */
optgzip = gt_option_new_bool("gzip", "write gzip compressed output file",
&ofi->gzip, false);
gt_option_parser_add_option(op, optgzip);
/* register option -bzip2 */
optbzip2 = gt_option_new_bool("bzip2", "write bzip2 compressed output file",
&ofi->bzip2, false);
gt_option_parser_add_option(op, optbzip2);
/* register option -force */
optforce = gt_option_new_bool(GT_FORCE_OPT_CSTR,
"force writing to output file",
&ofi->force, false);
gt_option_parser_add_option(op, optforce);
/* options -gzip and -bzip2 exclude each other */
gt_option_exclude(optgzip, optbzip2);
/* option implications */
gt_option_imply(optgzip, opto);
gt_option_imply(optbzip2, opto);
gt_option_imply(optforce, opto);
/* set hook function to determine <outfp> */
gt_option_parser_register_hook(op, determine_outfp, ofi);
}
static GtOptionParser* gt_compressedbits_option_parser_new(void *tool_arguments)
{
GtCompressdbitsArguments *arguments = tool_arguments;
GtOptionParser *op;
GtOption *option;
gt_assert(arguments);
/* init */
op = gt_option_parser_new("[option ...]",
"Testing compressed bitsequence, save to disk, reload.");
/* -size */
option = gt_option_new_ulong("size",
"size of GtBitsequence to create "
"(words 32/64 bit)",
&arguments->size, 20UL);
gt_option_parser_add_option(op, option);
arguments->size_op = gt_option_ref(option);
/* -samplerate */
option = gt_option_new_uint("samplerate",
"samplerate of random GtBitsequence to test",
&arguments->samplerate, 32U);
gt_option_parser_add_option(op, option);
/* -rand */
option = gt_option_new_bool("rand", "create random bitvector",
&arguments->fill_random, false);
gt_option_parser_add_option(op, option);
arguments->rand_op = gt_option_ref(option);
/* -check */
option = gt_option_new_bool("check", "compare original with compressed and "
"loaded from file",
&arguments->check_consistency, false);
gt_option_parser_add_option(op, option);
arguments->rand_op = gt_option_ref(option);
/* -input */
option = gt_option_new_filename(
"input",
"load vector from file, format is as follows:\n"
"[ULL size in bits][[ULL bits]...]\n"
" not usable with -size and -rand",
arguments->filename);
gt_option_parser_add_option(op, option);
arguments->filename_op = gt_option_ref(option);
gt_option_exclude(arguments->filename_op, arguments->size_op);
gt_option_exclude(arguments->filename_op, arguments->rand_op);
/* -benches */
option = gt_option_new_ulong("benches",
"number of function calls to benchmark",
&arguments->benches, 100000UL);
gt_option_parser_add_option(op, option);
return op;
}
static GtOptionParser* gt_speck_option_parser_new(void *tool_arguments)
{
GtOptionParser *op;
GtOption *option;
SpeccheckArguments *arguments = tool_arguments;
/* init */
op = gt_option_parser_new("[options] [GFF3_file ...]",
"Checks spec definition compliance in GFF3 input.");
option = gt_option_new_filename("specfile",
"file with specification definition",
arguments->specfile);
gt_option_parser_add_option(op, option);
gt_option_is_mandatory(option);
option = gt_option_new_bool("colored", "show colored output",
&arguments->colored, true);
gt_option_parser_add_option(op, option);
option = gt_option_new_bool("provideindex", "provide feature index in "
"specfile namespace (requires O(n) memory for n "
"input features)",
&arguments->provideindex, false);
gt_option_parser_add_option(op, option);
option = gt_option_new_bool("sort", "sort input before checking (requires "
"O(n) memory for n input features)",
&arguments->sort, false);
gt_option_parser_add_option(op, option);
option = gt_option_new_bool("failhard", "stop processing and report runtime "
"errors instead of recording them in the results",
&arguments->fail_hard, false);
gt_option_parser_add_option(op, option);
/* -format */
option = gt_option_new_string("output", "output format\n"
"choose from: json, text, html, statsonly",
arguments->format, "text");
gt_option_parser_add_option(op, option);
gt_typecheck_info_register_options_with_default(arguments->tci, op, "so");
gt_seqid2file_register_options_ext(op, arguments->s2fi, false, false);
gt_output_file_info_register_options(arguments->ofi, op, &arguments->outfp);
option = gt_option_new_verbose(&arguments->verbose);
gt_option_parser_add_option(op, option);
return op;
}
static GtOptionParser* gt_speck_option_parser_new(void *tool_arguments)
{
GtOptionParser *op;
GtOption *option;
SpeccheckArguments *arguments = tool_arguments;
/* init */
op = gt_option_parser_new("[options] [GFF3_file ...]",
"Checks spec definition compliance in GFF3 input.");
option = gt_option_new_filename("specfile",
"file with specification definition",
arguments->specfile);
gt_option_parser_add_option(op, option);
gt_option_is_mandatory(option);
option = gt_option_new_bool("colored", "show colored output",
&arguments->colored, true);
gt_option_parser_add_option(op, option);
option = gt_option_new_bool("allexpects", "show results counted by "
"expectations instead of by nodes",
&arguments->allexpects, false);
gt_option_parser_add_option(op, option);
option = gt_option_new_bool("provideindex", "provide feature index in "
"specfile namespace (requires O(n) memory for n "
"input features)",
&arguments->provideindex, false);
gt_option_parser_add_option(op, option);
option = gt_option_new_bool("sort", "sort input before checking (requires "
"O(n) memory for n input features)",
&arguments->sort, false);
gt_option_parser_add_option(op, option);
option = gt_option_new_bool("failhard", "stop processing and report runtime "
"errors instead of recording them in the results",
&arguments->fail_hard, false);
gt_option_parser_add_option(op, option);
option = gt_option_new_verbose(&arguments->verbose);
gt_option_parser_add_option(op, option);
gt_seqid2file_register_options(op, arguments->s2fi);
gt_output_file_info_register_options(arguments->ofi, op, &arguments->outfp);
return op;
}
static GtOptionParser*
gt_condenser_extract_option_parser_new(void *tool_arguments)
{
GtCondenserExtractArguments *arguments = tool_arguments;
GtOptionParser *op;
GtOption *option;
gt_assert(arguments);
/* init */
/*TODO soll das nur zu fasta sein oder kann man auch raw sequence
extracten?*/
op = gt_option_parser_new("[option ...] [archive]",
"Decompresses a condenser archive to fasta.");
/* -original */
option = gt_option_new_filename("original",
"uncompressed encseq, needs to be present "
"for development reasons.",
arguments->original);
gt_option_is_mandatory(option);
gt_option_parser_add_option(op, option);
/* -range */
option = gt_option_new_range("range",
"Range of positions to extract"
". If no "
"range is given, whole sequence "
"collection is extracted.",
&arguments->range, NULL);
gt_option_parser_add_option(op, option);
/* -verbose */
option = gt_option_new_bool("verbose", "Print out verbose output to stderr.",
&arguments->verbose, false);
gt_option_parser_add_option(op, option);
return op;
}
static GtOptionParser* gt_extractseq_option_parser_new(void *tool_arguments)
{
ExtractSeqArguments *arguments = tool_arguments;
GtOptionParser *op;
GtOption *frompos_option, *topos_option, *match_option, *width_option,
*fastakeyfile_option;
gt_assert(arguments);
/* init */
op = gt_option_parser_new("[option ...] [sequence_file(s)] | fastaindex",
"Extract sequences from given sequence file(s) or "
"fastaindex.");
/* -frompos */
frompos_option = gt_option_new_ulong_min(FROMPOS_OPTION_STR,
"extract sequence from this position\n"
"counting from 1 on",
&arguments->frompos, 0, 1UL);
gt_option_parser_add_option(op, frompos_option);
/* -topos */
topos_option = gt_option_new_ulong_min(TOPOS_OPTION_STR,
"extract sequence up to this position\n"
"counting from 1 on",
&arguments->topos, 0, 1UL);
gt_option_parser_add_option(op, topos_option);
/* -match */
match_option = gt_option_new_string("match", "extract all sequences whose "
"description matches the given pattern.\n"
"The given pattern must be a valid extended "
"regular expression.", arguments->pattern,
NULL);
gt_option_parser_add_option(op, match_option);
/* -keys */
fastakeyfile_option = gt_option_new_filename("keys",
"extract substrings for keys "
"in specified file",
arguments->fastakeyfile);
gt_option_parser_add_option(op, fastakeyfile_option);
/* -width */
width_option = gt_option_new_width(&arguments->width);
gt_option_parser_add_option(op, width_option);
/* output file options */
gt_outputfile_register_options(op, &arguments->outfp, arguments->ofi);
/* option implications */
gt_option_imply(frompos_option, topos_option);
gt_option_imply(topos_option, frompos_option);
/* option exclusions */
gt_option_exclude(frompos_option, match_option);
gt_option_exclude(topos_option, match_option);
gt_option_exclude(frompos_option, fastakeyfile_option);
gt_option_exclude(match_option, fastakeyfile_option);
gt_option_parser_set_comment_func(op, gt_gtdata_show_help, NULL);
return op;
}
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;
}
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_ulong_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_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;
}
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_extended_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;
}
static GtOPrval parse_options(int *parsed_args,
bool doesa,
Suffixeratoroptions *so,
int argc, const char **argv, GtError *err)
{
GtOptionParser *op;
GtOption *option,
*optionshowprogress,
*optiongenomediff,
*optionii;
GtOPrval oprval;
gt_error_check(err);
op = gt_option_parser_new("[option ...] (-db file [...] | -ii index)",
doesa ? "Compute enhanced suffix array."
: "Compute packed index.");
gt_option_parser_set_mail_address(op, "<*****@*****.**>");
/* input info */
so->indexname = gt_str_new();
so->inputindex = gt_str_new();
so->db = gt_str_array_new();
/* register options for encoded sequence handling */
so->encopts = gt_encseq_options_register_encoding(op, so->indexname, so->db);
so->loadopts = gt_encseq_options_register_loading(op, so->indexname);
/* register options for index handling */
if (doesa)
so->idxopts = gt_index_options_register_esa(op, so->encopts);
else
so->idxopts = gt_index_options_register_packedidx(op, so->indexname,
so->encopts);
/* verbosity */
option = gt_option_new_verbose(&so->beverbose);
gt_option_parser_add_option(op, option);
optionshowprogress = gt_option_new_bool("showprogress",
"show a progress bar",
&so->showprogress,
false);
gt_option_parser_add_option(op, optionshowprogress);
optionii = gt_option_new_filename("ii", "specify existing encoded sequence",
so->inputindex);
gt_option_parser_add_option(op, optionii);
gt_option_is_mandatory_either(gt_encseq_options_db_option(so->encopts),
optionii);
gt_option_exclude(gt_encseq_options_db_option(so->encopts), optionii);
gt_option_exclude(optionii, gt_encseq_options_smap_option(so->encopts));
gt_option_exclude(optionii, gt_encseq_options_dna_option(so->encopts));
gt_option_exclude(optionii, gt_encseq_options_protein_option(so->encopts));
gt_option_exclude(optionii, gt_encseq_options_plain_option(so->encopts));
gt_option_exclude(optionii, gt_encseq_options_sat_option(so->encopts));
optiongenomediff = gt_option_new_bool("genomediff",
"directly process the lcp intervals using "
"the genomediff algorithm (suffix array and "
"lcp-tables are not output)",
&so->genomediff,
false);
gt_option_is_extended_option(optiongenomediff);
if (gt_index_options_outsuftab_option(so->idxopts) != NULL) {
gt_option_exclude(optiongenomediff,
gt_index_options_outsuftab_option(so->idxopts));
}
gt_option_parser_add_option(op, optiongenomediff);
/* suffixerator and friends do not take arguments */
gt_option_parser_set_min_max_args(op, 0U, 0U);
oprval = gt_option_parser_parse(op, parsed_args, argc, argv, gt_versionfunc,
err);
if (gt_str_length(so->indexname) == 0UL) {
/* we do not have an indexname yet, so there was none given in the
-indexname option and it could not be derived from the input filenames.
So it must be in the -ii parameter. */
char *basenameptr;
basenameptr = gt_basename(gt_str_get(so->inputindex));
gt_str_set(so->indexname, basenameptr);
gt_free(basenameptr);
}
gt_option_parser_delete(op);
return oprval;
}
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;
}
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'", >r->interactive, false);
gt_option_hide_default(o);
gt_option_parser_add_option(op, o);
o = gt_option_new_bool("q", "suppress warnings", >r->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",
>_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", >r->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(>r->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",
>r->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", >r->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", >r->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;
}
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;
}
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_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;
}
