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_encseq_decode_option_parser_new(void *tool_arguments)
{
GtOptionParser *op;
GtOption *option,
*optionsep,
*optionseq,
*optionseqrange,
*optionmode;
GtEncseqDecodeArguments *arguments =
(GtEncseqDecodeArguments*) tool_arguments;
static const char *modes[] = {"fasta", "concat", NULL};
/* init */
op = gt_option_parser_new("(sequence_file|indexname)",
"Decode/extract encoded sequences.");
/* encseq options */
arguments->eopts = gt_encseq_options_register_loading(op, NULL);
gt_encseq_options_add_readmode_option(op, arguments->dir);
/* -singlechars */
option = gt_option_new_bool("singlechars",
"do not use a GtEncseqReader but access each "
"sequence character separately",
&arguments->singlechars,
false);
gt_option_is_extended_option(option);
gt_option_parser_add_option(op, option);
/* -seq */
optionseq = gt_option_new_uword("seq",
"extract sequence identified by its number",
&arguments->seq,
GT_UNDEF_UWORD);
gt_option_parser_add_option(op, optionseq);
/* -seqrange */
optionseqrange = gt_option_new_range("seqrange",
"extract multiple consecutive sequences",
&arguments->seqrng,
NULL);
gt_option_parser_add_option(op, optionseqrange);
gt_option_exclude(optionseqrange, optionseq);
/* -output */
optionmode = gt_option_new_choice("output",
"specify output format "
"(choose from fasta|concat)",
arguments->mode,
modes[0],
modes);
gt_option_parser_add_option(op, optionmode);
/* -range */
option = gt_option_new_range("range",
"concatenated range to extract "
"(implies '-output concat')",
&arguments->rng,
NULL);
gt_option_parser_add_option(op, option);
gt_option_imply(option, optionmode);
/* -sepchar */
optionsep = gt_option_new_string("sepchar",
"specify character to print as SEPARATOR",
arguments->sepchar, "|");
gt_option_parser_add_option(op, optionsep);
gt_option_imply(optionsep, optionmode);
gt_option_parser_set_min_max_args(op, 1, 1);
return op;
}
void gth_sa_filter_register_options(GtOptionParser *op, GthSAFilter *sa_filter,
bool extended_options)
{
GtOption *o;
gt_assert(sa_filter && op);
/* -minalignmentscore */
o = gt_option_new_double_min_max(MINALIGNMENTSCORE_OPT_CSTR, "set the "
"minimum alignment score for spliced "
"alignments to be included into the set of "
"spliced alignments",
&sa_filter->min_alignmentscore,
GTH_DEFAULT_MIN_ALIGNMENTSCORE,
GTH_DEFAULT_MIN_ALIGNMENTSCORE,
GTH_DEFAULT_MAX_ALIGNMENTSCORE);
if (extended_options)
gt_option_is_extended_option(o);
gt_option_parser_add_option(op, o);
/* -maxalignmentscore */
o = gt_option_new_double_min_max(MAXALIGNMENTSCORE_OPT_CSTR, "set the "
"maximum alignment score for spliced "
"alignments to be included into the set of "
"spliced alignments",
&sa_filter->max_alignmentscore,
GTH_DEFAULT_MAX_ALIGNMENTSCORE,
GTH_DEFAULT_MIN_ALIGNMENTSCORE,
GTH_DEFAULT_MAX_ALIGNMENTSCORE);
if (extended_options)
gt_option_is_extended_option(o);
gt_option_parser_add_option(op, o);
/* -mincoverage */
o = gt_option_new_double_min_max(MINCOVERAGE_OPT_CSTR, "set the minimum "
"coverage for spliced alignments to be "
"included into the set of spliced "
"alignments",
&sa_filter->min_coverage,
GTH_DEFAULT_MIN_COVERAGE,
GTH_DEFAULT_MIN_COVERAGE,
GTH_DEFAULT_MAX_COVERAGE);
if (extended_options)
gt_option_is_extended_option(o);
gt_option_parser_add_option(op, o);
/* -maxcoverage */
o = gt_option_new_double_min_max(MAXCOVERAGE_OPT_CSTR, "set the maximum "
"coverage for spliced alignments to be "
"included into the set of spliced "
"alignments",
&sa_filter->max_coverage,
GTH_DEFAULT_MAX_COVERAGE,
GTH_DEFAULT_MIN_COVERAGE,
GTH_DEFAULT_MAX_COVERAGE);
if (extended_options)
gt_option_is_extended_option(o);
gt_option_parser_add_option(op, o);
/* register hooks to check the arguments after the option parsing */
gt_option_parser_register_hook(op, sa_filter_check_arguments, sa_filter);
}
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;
}
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;
}
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;
}
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;
}
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_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_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 OPrval parse_options(int *parsed_args,
LTRharvestoptions *lo,
int argc, const char **argv, GtError *err)
{
GtOptionParser *op;
GtOption *optionindex,
*optionltrsearchseqrange,
*optionseed,
*optionminlenltr,
*optionmaxlenltr,
*optionmindistltr,
*optionmaxdistltr,
*optionmintsd,
*optionmaxtsd,
*optionsimilar,
*optionmotif,
*optionmotifmis,
*optionvic,
*optionoverlaps,
*optionxdrop,
*optionmat,
*optionmis,
*optionins,
*optiondel,
*optionv,
*optionoffset,
*optionlongoutput,
*optionout,
*optionoutinner,
*optiongff3;
OPrval oprval;
GtRange default_ltrsearchseqrange = {0,0};
unsigned int vicinityforcorrectboundaries;
static const char *overlaps[] = {
"best", /* the default */
"no",
"all",
NULL
};
gt_error_check(err);
op = gt_option_parser_new("[option ...] -index filenameindex",
"Predict LTR retrotransposons.");
/* -index */
lo->str_indexname = gt_str_new();
optionindex = gt_option_new_string("index",
"specify the name of the enhanced suffix "
"array index (mandatory)",
lo->str_indexname, NULL);
gt_option_is_mandatory(optionindex);
gt_option_parser_add_option(op, optionindex);
/* -range */
optionltrsearchseqrange
= gt_option_new_range("range",
"specify sequence range in which LTRs are searched",
&lo->repeatinfo.ltrsearchseqrange,
&default_ltrsearchseqrange);
gt_option_parser_add_option(op, optionltrsearchseqrange);
/* -seed */
optionseed = gt_option_new_ulong_min("seed",
"specify minimum seed length for"
" exact repeats",
&lo->minseedlength,
30UL,
1UL);
gt_option_parser_add_option(op, optionseed);
/* -minlenltr */
optionminlenltr = gt_option_new_ulong_min_max("minlenltr",
"specify minimum length for each LTR",
&lo->repeatinfo.lmin,
100UL,
1UL,
GT_UNDEF_ULONG);
gt_option_parser_add_option(op, optionminlenltr);
/* -maxlenltr */
optionmaxlenltr = gt_option_new_ulong_min_max("maxlenltr",
"specify maximum length for each LTR",
&lo->repeatinfo.lmax,
1000UL,
1UL,
GT_UNDEF_ULONG);
gt_option_parser_add_option(op, optionmaxlenltr);
/* -mindistltr */
optionmindistltr = gt_option_new_ulong_min_max("mindistltr",
"specify minimum distance of "
"LTR startpositions",
&lo->repeatinfo.dmin,
1000UL,
1UL,
GT_UNDEF_ULONG);
gt_option_parser_add_option(op, optionmindistltr);
/* -maxdistltr */
optionmaxdistltr = gt_option_new_ulong_min_max("maxdistltr",
"specify maximum distance of "
"LTR startpositions",
&lo->repeatinfo.dmax,
15000UL,
1UL,
GT_UNDEF_ULONG);
gt_option_parser_add_option(op, optionmaxdistltr);
/* -similar */
optionsimilar = gt_option_new_double_min_max("similar",
"specify similaritythreshold in "
"range [1..100%]",
&lo->similaritythreshold,
(double) 85.0,
(double) 0.0,
100.0);
gt_option_parser_add_option(op, optionsimilar);
/* -mintsd */
optionmintsd = gt_option_new_uint_min_max("mintsd",
"specify minimum length for each TSD",
&lo->minlengthTSD,
4U,
0,
GT_UNDEF_UINT);
gt_option_parser_add_option(op, optionmintsd);
/* -maxtsd */
optionmaxtsd = gt_option_new_uint_min_max("maxtsd",
"specify maximum length for each TSD",
&lo->maxlengthTSD,
20U,
0,
GT_UNDEF_UINT);
gt_option_parser_add_option(op, optionmaxtsd);
/* -motif */
/* characters will be tranformed later
into characters from virtualtree alphabet */
lo->motif.firstleft = (GtUchar) 't';
lo->motif.secondleft = (GtUchar) 'g';
lo->motif.firstright = (GtUchar) 'c';
lo->motif.secondright = (GtUchar) 'a';
lo->motif.str_motif = gt_str_new();
optionmotif = gt_option_new_string("motif",
"specify 2 nucleotides startmotif + "
"2 nucleotides endmotif: ****",
lo->motif.str_motif, NULL);
gt_option_parser_add_option(op, optionmotif);
/* -motifmis */
optionmotifmis = gt_option_new_uint_min_max("motifmis",
"specify maximum number of "
"mismatches in motif [0,3]",
&lo->motif.allowedmismatches,
4U,
0,
3U);
gt_option_parser_add_option(op, optionmotifmis);
/* -vic */
optionvic = gt_option_new_uint_min_max("vic",
"specify the number of nucleotides (to the left and "
"to the right) that will be searched "
"for TSDs and/or motifs around 5' and 3' boundary "
"of predicted LTR retrotransposons",
&vicinityforcorrectboundaries,
60U,
1U,
500U);
gt_option_parser_add_option(op, optionvic);
/* -overlaps */
lo->str_overlaps = gt_str_new();
optionoverlaps = gt_option_new_choice("overlaps",
"specify no|best|all",
lo->str_overlaps,
overlaps[0],
overlaps);
gt_option_parser_add_option(op, optionoverlaps);
/* -xdrop */
optionxdrop = gt_option_new_int_min("xdrop",
"specify xdropbelowscore for extension-alignment",
&lo->xdropbelowscore,
(int)5,
(int)0);
gt_option_parser_add_option(op, optionxdrop);
/* -mat */
lo->arbitscores.gcd = 1; /* set only for initialization,
do not change! */
optionmat = gt_option_new_int_min("mat",
"specify matchscore for extension-alignment",
&lo->arbitscores.mat,
2,
1);
gt_option_parser_add_option(op, optionmat);
/* -mis */
optionmis = gt_option_new_int_max("mis",
"specify mismatchscore for extension-alignment",
&lo->arbitscores.mis,
-2,
-1);
gt_option_parser_add_option(op, optionmis);
/* -ins */
optionins = gt_option_new_int_max("ins",
"specify insertionscore for extension-alignment",
&lo->arbitscores.ins,
-3,
-1);
gt_option_parser_add_option(op, optionins);
/* -del */
optiondel = gt_option_new_int_max("del",
"specify deletionscore for extension-alignment",
&lo->arbitscores.del,
-3,
-1);
gt_option_parser_add_option(op, optiondel);
/* -v */
optionv = gt_option_new_bool("v",
"verbose mode",
&lo->verbosemode,
false);
gt_option_parser_add_option(op, optionv);
/* -longoutput */
optionlongoutput = gt_option_new_bool("longoutput",
"additional motif/TSD output",
&lo->longoutput,
false);
gt_option_parser_add_option(op, optionlongoutput);
/* -out */
lo->fastaoutput = false; /* by default no FASTA output */
lo->str_fastaoutputfilename = gt_str_new();
optionout = gt_option_new_string("out",
"specify FASTA outputfilename",
lo->str_fastaoutputfilename, NULL);
gt_option_parser_add_option(op, optionout);
/* -outinner */
lo->fastaoutputinnerregion = false;
lo->str_fastaoutputfilenameinnerregion = gt_str_new();
optionoutinner = gt_option_new_string("outinner",
"specify FASTA outputfilename for inner regions",
lo->str_fastaoutputfilenameinnerregion, NULL);
gt_option_parser_add_option(op, optionoutinner);
/* -gff3 */
lo->gff3output = false; /* by default no gff3 output */
lo->str_gff3filename = gt_str_new();
optiongff3 = gt_option_new_string("gff3",
"specify GFF3 outputfilename",
lo->str_gff3filename, NULL);
gt_option_parser_add_option(op, optiongff3);
/* -offset */
optionoffset = gt_option_new_ulong("offset",
"offset added to GFF3 coordinates",
&lo->offset,
0UL);
gt_option_parser_add_option(op, optionoffset);
gt_option_is_extended_option(optionoffset);
/* implications */
gt_option_imply(optionmaxtsd, optionmintsd);
gt_option_imply(optionmotifmis, optionmotif);
gt_option_imply_either_2(optionlongoutput, optionmintsd, optionmotif);
gt_option_parser_refer_to_manual(op);
oprval = gt_option_parser_parse(op, parsed_args, argc, argv, gt_versionfunc,
err);
lo->vicinityforcorrectboundaries = (Seqpos) vicinityforcorrectboundaries;
if (oprval == OPTIONPARSER_OK)
{
if (lo->repeatinfo.lmin > lo->repeatinfo.lmax)
{
gt_error_set(err,"argument of -minlenltr is greater than argument of"
" -maxlenltr");
oprval = OPTIONPARSER_ERROR;
}
if (lo->repeatinfo.dmin > lo->repeatinfo.dmax)
{
gt_error_set(err,
"argument of -mindistltr is greater than argument of -maxdistltr");
oprval = OPTIONPARSER_ERROR;
}
if (lo->repeatinfo.lmax > lo->repeatinfo.dmin)
{
gt_error_set(err,"argument of -maxlenltr is greater than argument of"
" -mindistltr");
oprval = OPTIONPARSER_ERROR;
}
if (lo->minlengthTSD > lo->maxlengthTSD)
{
gt_error_set(err,
"argument of -mintsd is greater than argument of -maxtsd");
oprval = OPTIONPARSER_ERROR;
}
/* If option motif is set,
store characters, transform them later */
if (gt_option_is_set(optionmotif))
{
if (gt_str_length(lo->motif.str_motif) != 4UL)
{
gt_error_set(err,
"argument of -motif has not exactly 4 characters");
oprval = OPTIONPARSER_ERROR;
}
lo->motif.firstleft = (GtUchar) gt_str_get(lo->motif.str_motif)[0];
lo->motif.secondleft = (GtUchar) gt_str_get(lo->motif.str_motif)[1];
lo->motif.firstright = (GtUchar) gt_str_get(lo->motif.str_motif)[2];
lo->motif.secondright = (GtUchar) gt_str_get(lo->motif.str_motif)[3];
/* default if motif specified */
if (!gt_option_is_set(optionmotifmis))
{
lo->motif.allowedmismatches = 0;
}
}
/* If option overlaps is set */
if (gt_option_is_set(optionoverlaps))
{
if (strcmp(gt_str_get(lo->str_overlaps), "no") == 0)
{
lo->bestofoverlap = false;
lo->nooverlapallowed = true;
}
else if (strcmp(gt_str_get(lo->str_overlaps), "best") == 0 )
{
lo->bestofoverlap = true;
lo->nooverlapallowed = false;
}
else if (strcmp(gt_str_get(lo->str_overlaps), "all") == 0 )
{
lo->bestofoverlap = false;
lo->nooverlapallowed = false;
}
else
{
gt_assert(0); /* cannot happen */
}
}
else
{
/* default is "best" */
lo->bestofoverlap = true; /* take best prediction
if overlap occurs, default */
lo->nooverlapallowed = false; /* overlapping predictions (not)allowed*/
}
/* if FASTA output is set */
if (gt_option_is_set(optionout))
{
lo->fastaoutput = true;
}
/* if FASTA output inner region is set */
if (gt_option_is_set(optionoutinner))
{
lo->fastaoutputinnerregion = true;
}
/* if GFF3 output is set */
if (gt_option_is_set(optiongff3))
{
lo->gff3output = true;
}
if (gt_option_is_set(optionltrsearchseqrange))
{
if (lo->repeatinfo.ltrsearchseqrange.start >
lo->repeatinfo.ltrsearchseqrange.end)
{
gt_error_set(err,
"arguments of -range: first arg must be <= than second arg");
oprval = OPTIONPARSER_ERROR;
}
}
}
gt_option_parser_delete(op);
return oprval;
}
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;
}
