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_sequniq_option_parser_new(void *tool_arguments)
{
GtSequniqArguments *arguments = tool_arguments;
GtOptionParser *op;
GtOption *seqit_option, *verbose_option, *width_option, *rev_option,
*nofseqs_option;
gt_assert(arguments);
op = gt_option_parser_new("[option ...] sequence_file [...] ",
"Filter out repeated sequences in given "
"sequence files.");
/* -seqit */
seqit_option = gt_option_new_bool("seqit", "use sequence iterator",
&arguments->seqit, false);
gt_option_is_development_option(seqit_option);
gt_option_parser_add_option(op, seqit_option);
/* -nofseqs */
nofseqs_option = gt_option_new_uword("nofseqs", "number of sequences "
"(improves efficiency)\ndefault: unspecified",
&arguments->nofseqs, 0);
gt_option_is_development_option(nofseqs_option);
gt_option_hide_default(nofseqs_option);
gt_option_parser_add_option(op, nofseqs_option);
/* -rev */
rev_option = gt_option_new_bool("rev", "also filter out sequences whose "
"reverse complement is identical to a sequence already output",
&arguments->rev, false);
gt_option_parser_add_option(op, rev_option);
/* -v */
verbose_option = gt_option_new_verbose(&arguments->verbose);
gt_option_parser_add_option(op, verbose_option);
/* -width */
width_option = gt_option_new_width(&arguments->width);
gt_option_parser_add_option(op, width_option);
gt_output_file_info_register_options(arguments->ofi, op, &arguments->outfp);
/* option implications */
gt_option_imply(verbose_option, seqit_option);
gt_option_parser_set_comment_func(op, gt_gtdata_show_help, NULL);
gt_option_parser_set_min_args(op, 1U);
return op;
}
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_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_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_select_option_parser_new(void *tool_arguments)
{
SelectArguments *arguments = tool_arguments;
GtOptionParser *op;
GtOption *option, *contain_option, *overlap_option, *minaveragessp_option,
*singleintron_option, *optiondroppedfile;
gt_assert(arguments);
static const char *filter_logic[] = {
"AND",
"OR",
NULL
};
/* init */
op = gt_option_parser_new("[option ...] [GFF3_file ...]",
"Select certain features (specified by the used "
"options) from given GFF3 file(s).");
/* -seqid */
option = gt_option_new_string("seqid", "select feature with the given "
"sequence ID (all comments are selected). ",
arguments->seqid, NULL);
gt_option_parser_add_option(op, option);
/* -source */
option = gt_option_new_string("source", "select feature with the given "
"source (the source is column 2 in regular "
"GFF3 lines)" , arguments->source, NULL);
gt_option_parser_add_option(op, option);
/* -contain */
contain_option = gt_option_new_range("contain", "select all features which "
"are contained in the given range",
&arguments->contain_range, NULL);
gt_option_parser_add_option(op, contain_option);
/* -overlap */
overlap_option = gt_option_new_range("overlap", "select all features which "
"do overlap with the given range",
&arguments->overlap_range, NULL);
gt_option_parser_add_option(op, overlap_option);
/* -strand */
option = gt_option_new_string(GT_STRAND_OPT, "select all top-level features"
"(i.e., features without parents) whose strand "
"equals the given one (must be one of '"
GT_STRAND_CHARS"')", arguments->gt_strand_char,
NULL);
gt_option_parser_add_option(op, option);
/* -targetstrand */
option = gt_option_new_string(TARGETGT_STRAND_OPT, "select all top-level "
"features (i.e., features without parents) "
"which have exactly one target attribute whose "
"strand equals the given one (must be one of '"
GT_STRAND_CHARS"')",
arguments->targetgt_strand_char, NULL);
gt_option_parser_add_option(op, option);
/* -targetbest */
option = gt_option_new_bool("targetbest", "if multiple top-level features "
"(i.e., features without parents) with exactly "
"one target attribute have the same target_id, "
"keep only the feature with the best score. If "
"-"TARGETGT_STRAND_OPT" is used at the same time, "
"this option is applied after "
"-"TARGETGT_STRAND_OPT".\n"
"Memory consumption is proportional to the input "
"file size(s).", &arguments->targetbest, false);
gt_option_parser_add_option(op, option);
/* -hascds */
option = gt_option_new_bool("hascds", "select all top-level features which "
"do have a CDS child", &arguments->has_CDS,
false);
gt_option_parser_add_option(op, option);
/* -maxgenelength */
option = gt_option_new_uword_min("maxgenelength", "select genes up to the "
"given maximum length",
&arguments->max_gene_length, GT_UNDEF_UWORD,
1);
gt_option_parser_add_option(op, option);
/* -maxgenenum */
option = gt_option_new_uword("maxgenenum", "select the first genes up to the "
"given maximum number", &arguments->max_gene_num,
GT_UNDEF_UWORD);
gt_option_parser_add_option(op, option);
/* -mingenescore */
option = gt_option_new_double("mingenescore", "select genes with the given "
"minimum score", &arguments->min_gene_score,
GT_UNDEF_DOUBLE);
gt_option_parser_add_option(op, option);
/* -maxgenescore */
option = gt_option_new_double("maxgenescore", "select genes with the given "
"maximum score", &arguments->max_gene_score,
GT_UNDEF_DOUBLE);
gt_option_parser_add_option(op, option);
/* -minaveragessp */
minaveragessp_option =
gt_option_new_probability("minaveragessp",
"set the minimum average splice site probability",
&arguments->min_average_splice_site_prob,
GT_UNDEF_DOUBLE);
gt_option_parser_add_option(op, minaveragessp_option);
/* -singleintronfactor */
singleintron_option =
gt_option_new_double_min("singleintronfactor",
"factor to multiplicate the average splice site "
"probability with for single introns before "
"comparing it to the minimum average splice site "
"probability", &arguments->single_intron_factor,
1.0, 1.0);
gt_option_is_development_option(singleintron_option);
gt_option_parser_add_option(op, singleintron_option);
/* -featurenum */
option = gt_option_new_uword_min("featurenum",
"select feature tree occurring "
"at given position in input",
&arguments->feature_num, GT_UNDEF_UWORD, 1);
gt_option_is_development_option(option);
gt_option_parser_add_option(op, option);
/* -filter_files */
option = gt_option_new_filename_array("rule_files",
"specify Lua files to be used "
"for selection",
arguments->filter_files);
gt_option_parser_add_option(op, option);
/* -filter_logic */
option = gt_option_new_choice("rule_logic", "select how multiple Lua "
"files should be combined\nchoose from AND|OR",
arguments->filter_logic, filter_logic[0],
filter_logic);
gt_option_parser_add_option(op, option);
/* -nh_file */
optiondroppedfile = gt_option_new_filename("dropped_file",
"save non-selected features to "
"file",
arguments->dropped_file);
gt_option_parser_add_option(op, optiondroppedfile);
/* -v */
option = gt_option_new_verbose(&arguments->verbose);
gt_option_parser_add_option(op, option);
/* option exclusions */
gt_option_exclude(contain_option, overlap_option);
/* option implications */
gt_option_imply(singleintron_option, minaveragessp_option);
/* output file options */
gt_output_file_info_register_options(arguments->ofi, op, &arguments->outfp);
gt_option_parser_set_comment_func(op, gt_gtdata_show_help, NULL);
return op;
}
static GtOPrval parse_options(int *parsed_args,
Cmppairwiseopt *pw,
int argc, const char **argv, GtError *err)
{
GtOptionParser *op;
GtOption *optionstrings,
*optionfiles,
*optioncharlistlen,
*optiontext,
*optionshowedist;
GtStrArray *charlistlen;
GtOPrval oprval;
gt_error_check(err);
charlistlen = gt_str_array_new();
pw->strings = gt_str_array_new();
pw->files = gt_str_array_new();
pw->text = gt_str_new();
pw->charlistlen = NULL;
pw->showedist = false;
op = gt_option_parser_new("options", "Apply function to pairs of strings.");
gt_option_parser_set_mail_address(op, "<*****@*****.**>");
optionstrings = gt_option_new_string_array("ss", "use two strings",
pw->strings);
gt_option_parser_add_option(op, optionstrings);
optionfiles = gt_option_new_filename_array("ff", "use two files",
pw->files);
gt_option_parser_add_option(op, optionfiles);
optioncharlistlen = gt_option_new_string_array("a",
"use character list and length",
charlistlen);
gt_option_parser_add_option(op, optioncharlistlen);
optiontext = gt_option_new_string("t", "use text", pw->text, NULL);
gt_option_parser_add_option(op, optiontext);
optionshowedist = gt_option_new_bool("e", "output unit edit distance",
&pw->showedist, false);
gt_option_parser_add_option(op, optionshowedist);
gt_option_exclude(optionstrings, optionfiles);
gt_option_exclude(optionstrings, optioncharlistlen);
gt_option_exclude(optionstrings, optiontext);
gt_option_exclude(optionfiles, optioncharlistlen);
gt_option_exclude(optionfiles, optiontext);
gt_option_exclude(optioncharlistlen, optiontext);
gt_option_imply(optionshowedist, optionstrings);
oprval = gt_option_parser_parse(op, parsed_args, argc, argv, gt_versionfunc,
err);
if (oprval == GT_OPTION_PARSER_OK)
{
if (gt_option_is_set(optionstrings))
{
if (gt_str_array_size(pw->strings) != 2UL)
{
gt_error_set(err, "option -ss requires two string arguments");
oprval = GT_OPTION_PARSER_ERROR;
}
} else
{
if (gt_option_is_set(optionfiles))
{
if (gt_str_array_size(pw->files) != 2UL)
{
gt_error_set(err, "option -ff requires two filename arguments");
oprval = GT_OPTION_PARSER_ERROR;
}
} else
{
if (gt_option_is_set(optioncharlistlen))
{
GtWord readint;
if (gt_str_array_size(charlistlen) != 2UL)
{
gt_error_set(err,
"option -a requires charlist and length argument");
oprval = GT_OPTION_PARSER_ERROR;
}
pw->charlistlen = gt_malloc(sizeof *pw->charlistlen);
pw->charlistlen->charlist =
gt_str_ref(gt_str_array_get_str(charlistlen,
0));
if (sscanf(gt_str_array_get(charlistlen,1UL), GT_WD, &readint) != 1 ||
readint < 1L)
{
gt_error_set(err,
"option -a requires charlist and length argument");
oprval = GT_OPTION_PARSER_ERROR;
}
pw->charlistlen->len = (GtUword) readint;
} else
{
if (!gt_option_is_set(optiontext))
{
gt_error_set(err,
"use exactly one of the options -ss, -ff, -a, -t");
oprval = GT_OPTION_PARSER_ERROR;
}
}
}
}
}
gt_option_parser_delete(op);
if (oprval == GT_OPTION_PARSER_OK && *parsed_args != argc)
{
gt_error_set(err, "superfluous program parameters");
oprval = GT_OPTION_PARSER_ERROR;
}
gt_str_array_delete(charlistlen);
return oprval;
}
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_kmer_database_option_parser_new(void *tool_arguments)
{
GtKmerDatabaseArguments *arguments = tool_arguments;
GtOptionParser *op;
GtOption *option,
*option_verbose,
*option_use_cutoff,
*option_hash,
*option_mean_cutoff;
gt_assert(arguments);
/* init */
op = gt_option_parser_new("[option ...] [file]",
"Makes a GtKmerDatabase from the input file.");
/* -kmersize */
option = gt_option_new_uint_min_max("kmersize", "kmersize used",
&arguments->kmersize, 3U, 1U, 10U);
gt_option_parser_add_option(op, option);
/* -verbose */
option_verbose = gt_option_new_bool("verbose", "prints out results of "
"merging",
&arguments->verbose, false);
gt_option_parser_add_option(op, option_verbose);
/* -merge_only */
option = gt_option_new_bool("merge_only", "only uses merge to build DB, "
"doesn_t build two DBs to compare merge with a "
"different method (much faster). It also allows "
"for random intervals which are biffer than the "
"maximum buffer size (will be split internally).",
&arguments->merge_only, false);
gt_option_parser_add_option(op, option);
/* -use_cutoff */
option_use_cutoff = gt_option_new_bool("use_cutoff", "uses a cutoff. see "
"-set_cutoff description. Only works "
"with merge_only",
&arguments->cutoff, false);
gt_option_parser_add_option(op, option_use_cutoff);
gt_option_imply(option_use_cutoff, option);
/* -set_cutoff */
option = gt_option_new_uword_min("set_cutoff", "kmers occuring more often "
"than this value won't be saved",
&arguments->cutoff_value, (GtUword) 30,
(GtUword) 1);
gt_option_parser_add_option(op, option);
gt_option_imply(option, option_use_cutoff);
/* -mean_cutoff */
option_mean_cutoff = gt_option_new_bool("mean_cutoff", "2*mean of kmer "
"occurrence will be"
" used as cutoff value",
&arguments->mean_cutoff, false);
gt_option_parser_add_option(op, option_mean_cutoff);
gt_option_imply(option_mean_cutoff, option_use_cutoff);
gt_option_exclude(option_mean_cutoff, option);
/* -disable_prune */
option = gt_option_new_bool("disable_prune", "disables the removel of kmers, "
"which occure more often than the cutoff.",
&arguments->prune, false);
gt_option_parser_add_option(op, option);
gt_option_imply(option, option_use_cutoff);
/* -use_hash */
option_hash = gt_option_new_bool("use_hash", "saves each kmer in kdb and "
"also in a hash. afterwards both will be "
"accessed and time for saving and "
"accessing will be shown.",
&arguments->use_hash, false);
gt_option_parser_add_option(op, option_hash);
gt_option_exclude(option_hash, option_use_cutoff);
gt_option_exclude(option_hash, option_verbose);
/* -benchmark */
option = gt_option_new_bool("benchmark", "measures the time the tool takes to"
" fill the database. Doesn't test for consistency"
" though!", &arguments->bench, false);
gt_option_parser_add_option(op, option);
/* -bsize */
option = gt_option_new_uword_min("bsize", "size of the buffer",
&arguments->sb_size, (GtUword) 100000,
(GtUword) 2);
gt_option_parser_add_option(op, option);
gt_option_exclude(option, option_hash);
/* -outfile */
option = gt_option_new_string("outfile", "specifies file for verbose output",
arguments->print_filename, NULL);
gt_option_parser_add_option(op, option);
return op;
}
static GtOptionParser* gt_tir_option_parser_new(void *tool_arguments)
{
GtTirArguments *arguments = tool_arguments;
GtOptionParser *op;
GtOption *optionindex, /* index */
*optionseed, /* minseedlength */
*optionminlentir, /* minimal length of TIR */
*optionmaxlentir, /* maximal length of TIR */
*optionmindisttir, /* minimal distance of TIRs */
*optionmaxdisttir, /* maximal distance of TIRs */
*optionmat, /* arbitrary scores */
*optionmis,
*optionins,
*optiondel,
*optionxdrop, /* xdropbelowscore for extension alignment */
*optionsimilar, /* similarity threshold */
*optionoverlaps, /* for overlaps */
*optionmintsd, /* minimal length for Target Site Duplication */
*optionmaxtsd, /* maximal length for Target Site Duplication */
*optionvicinity, /* vicinity around TIRs to be searched for TSDs */
*optionhmms,
*optionevalcutoff,
*optionpdomcutoff,
*optionmaxgap;
static const char *overlaps[] = {
"best", /* default */
"no",
"all",
NULL
};
static const char *cutoffs[] = {
"NONE",
"GA",
"TC",
NULL
};
gt_assert(arguments);
/* init */
op = gt_option_parser_new("[option ...] -index INDEXNAME",
"Identify Terminal Inverted Repeat (TIR) elements,"
"such as DNA transposons.");
/* -index */
optionindex = gt_option_new_string("index",
"specify the name of the enhanced suffix "
"array index (mandatory)",
arguments->str_indexname, NULL);
gt_option_is_mandatory(optionindex);
gt_option_parser_add_option(op, optionindex);
/* -seed */
optionseed = gt_option_new_uword_min("seed",
"specify minimum seed length for "
"exact repeats",
&arguments->min_seed_length, 20UL, 2UL);
gt_option_parser_add_option(op, optionseed);
/* -minlentir */
optionminlentir = gt_option_new_uword_min_max("mintirlen",
"specify minimum length for "
"each TIR",
&arguments->min_TIR_length,
27UL, 1UL, GT_UNDEF_UWORD);
gt_option_parser_add_option(op, optionminlentir);
/* -maxlentir */
optionmaxlentir = gt_option_new_uword_min_max("maxtirlen",
"specify maximum length for "
"each TIR",
&arguments->max_TIR_length,
1000UL, 1UL, GT_UNDEF_UWORD);
gt_option_parser_add_option(op, optionmaxlentir);
/* -mindisttir */
optionmindisttir = gt_option_new_uword_min_max("mintirdist",
"specify minimum distance of "
"TIRs",
&arguments->min_TIR_distance,
100UL, 1UL, GT_UNDEF_UWORD);
gt_option_parser_add_option(op, optionmindisttir);
/* -maxdisttir */
optionmaxdisttir = gt_option_new_uword_min_max("maxtirdist",
"specify maximum distance of "
"TIRs",
&arguments->max_TIR_distance,
10000UL, 1UL, GT_UNDEF_UWORD);
gt_option_parser_add_option(op, optionmaxdisttir);
optionmat = gt_option_new_int_min("mat",
"specify matchscore for "
"extension-alignment",
&arguments->arbit_scores.mat, 2, 1);
gt_option_parser_add_option(op, optionmat);
/* -mis */
optionmis = gt_option_new_int_max("mis",
"specify mismatchscore for "
"extension-alignment",
&arguments->arbit_scores.mis, -2, -1);
gt_option_parser_add_option(op, optionmis);
/* -ins */
optionins = gt_option_new_int_max("ins",
"specify insertionscore for "
"extension-alignment",
&arguments->arbit_scores.ins, -3, -1);
gt_option_parser_add_option(op, optionins);
/* -del */
optiondel = gt_option_new_int_max("del",
"specify deletionscore for "
"extension-alignment",
&arguments->arbit_scores.del, -3, -1);
gt_option_parser_add_option(op, optiondel);
/* -xdrop */
optionxdrop = gt_option_new_int_min("xdrop",
"specify xdropbelowscore for "
"extension-alignment",
&arguments->xdrop_belowscore, (int) 5,
(int) 0);
gt_option_parser_add_option(op, optionxdrop);
/* -similar */
optionsimilar = gt_option_new_double_min_max("similar",
"specify similaritythreshold in "
"range [1..100%]",
&arguments->similarity_threshold,
(double) 85.0, (double) 0.0,
100.0);
gt_option_parser_add_option(op, optionsimilar);
/* -overlaps */
optionoverlaps = gt_option_new_choice("overlaps", "specify no|best|all",
arguments->str_overlaps,
overlaps[0], overlaps);
gt_option_parser_add_option(op, optionoverlaps);
arguments->optionoverlaps = gt_option_ref(optionoverlaps);
/* -mintsd */
optionmintsd = gt_option_new_uword_min_max("mintsd",
"specify minimum length for each "
"TSD",
&arguments->min_TSD_length,
2U, 0, GT_UNDEF_UINT);
gt_option_parser_add_option(op, optionmintsd);
/* -maxtsd */
optionmaxtsd = gt_option_new_uword_min_max("maxtsd",
"specify maximum length for each "
"TSD",
&arguments->max_TSD_length,
11U, 0, GT_UNDEF_UINT);
gt_option_parser_add_option(op, optionmaxtsd);
gt_option_imply(optionmaxtsd, optionmintsd);
/* -vicinity */
optionvicinity = gt_option_new_uword_min_max("vic",
"specify the number of "
"nucleotides (to the left and "
"to the right) that will be "
"searched for TSDs around 5' "
"and 3' boundary of predicted "
"TIRs",
&arguments->vicinity,
60U, 1U, 500U);
gt_option_parser_add_option(op, optionvicinity);
optionhmms = gt_option_new_filename_array("hmms",
"profile HMM models for domain detection "
"(separate by spaces, finish with --) in "
"HMMER3 format\n"
"Omit this option to disable pHMM search.",
arguments->hmm_files);
gt_option_parser_add_option(op, optionhmms);
optionevalcutoff = gt_option_new_probability("pdomevalcutoff",
"global E-value cutoff for pHMM search\n"
"default 1E-6",
&arguments->evalue_cutoff,
0.000001);
gt_option_parser_add_option(op, optionevalcutoff);
gt_option_hide_default(optionevalcutoff);
gt_option_imply(optionevalcutoff, optionhmms);
optionpdomcutoff = gt_option_new_choice("pdomcutoff",
"model-specific score cutoff\n"
"choose from TC (trusted cutoff) | "
"GA (gathering cutoff) | "
"NONE (no cutoffs)",
arguments->cutoffs, cutoffs[1], cutoffs);
gt_option_parser_add_option(op, optionpdomcutoff);
gt_option_imply(optionpdomcutoff, optionhmms);
/* XXX: make -pdomcutoff and -pdomevalcutoff mutually exclusive */
optionmaxgap = gt_option_new_uint("maxgaplen",
"maximal allowed gap size between fragments (in amino "
"acids) when chaining pHMM hits for a protein domain",
&arguments->chain_max_gap_length,
50U);
gt_option_parser_add_option(op, optionmaxgap);
gt_option_is_extended_option(optionmaxgap);
gt_option_imply(optionmaxgap, optionhmms);
return op;
}
static GtOptionParser* gt_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 GtOptionParser *gt_tyr_mkindex_option_parser_new(void *tool_arguments)
{
GtOptionParser *op;
GtOption *option,
*optionminocc,
*optionmaxocc,
*optionpl,
*optionstoreindex,
*optionstorecounts,
*optionscan,
*optionesa;
Tyr_mkindex_options *arguments = tool_arguments;
op = gt_option_parser_new("[options] -esa suffixerator-index [options]",
"Count and index k-mers in the given enhanced "
"suffix array for a fixed value of k.");
gt_option_parser_set_mailaddress(op,"<*****@*****.**>");
optionesa = gt_option_new_string("esa","specify suffixerator-index\n"
"(mandatory option)",
arguments->str_inputindex,
NULL);
gt_option_is_mandatory(optionesa);
gt_option_parser_add_option(op, optionesa);
option = gt_option_new_ulong("mersize",
"Specify the mer size.",
&arguments->mersize,
20UL);
gt_option_parser_add_option(op, option);
optionminocc
= gt_option_new_ulong("minocc",
"Specify the minimum occurrence number for "
"the mers to output/index",
&arguments->userdefinedminocc,
0);
gt_option_parser_add_option(op, optionminocc);
optionmaxocc
= gt_option_new_ulong("maxocc",
"Specify the maximum occurrence number for "
"the mers to output/index",
&arguments->userdefinedmaxocc,
0);
gt_option_parser_add_option(op, optionmaxocc);
optionpl = gt_option_new_uint_min("pl",
"specify prefix length for bucket boundary construction\n"
"recommendation: use without argument;\n"
"then a reasonable prefix length is automatically determined",
&arguments->userdefinedprefixlength,
0,
1U);
gt_option_argument_is_optional(optionpl);
gt_option_parser_add_option(op, optionpl);
arguments->refoptionpl = gt_option_ref(optionpl);
optionstoreindex = gt_option_new_string("indexname",
"store the mers specified by options "
"-maxocc and -minocc in an index",
arguments->str_storeindex, NULL);
gt_option_parser_add_option(op, optionstoreindex);
optionstorecounts = gt_option_new_bool("counts", "store counts of the mers",
&arguments->storecounts,false);
gt_option_parser_add_option(op, optionstorecounts);
option = gt_option_new_bool("test", "perform tests to verify program "
"correctness", &arguments->performtest,
false);
gt_option_is_development_option(option);
gt_option_parser_add_option(op, option);
optionscan = gt_option_new_bool("scan",
"read enhanced suffix array sequentially "
"instead of mapping it to memory",
&arguments->scanfile,
false);
gt_option_parser_add_option(op, optionscan);
option = gt_option_new_verbose(&arguments->verbose);
gt_option_parser_add_option(op, option);
gt_option_imply(optionpl, optionstoreindex);
gt_option_imply(optionstorecounts, optionstoreindex);
gt_option_imply_either_2(optionstoreindex,optionminocc,optionmaxocc);
return op;
}
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_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_condenseq_extract_option_parser_new(void *tool_arguments)
{
GtCondenserExtractArguments *arguments = tool_arguments;
GtOptionParser *op;
GtOption *option,
*optionrange,
*optionseq,
*optionseqrange,
*optionmode;
static const char *modes[] = {"fasta", "concat", NULL};
gt_assert(arguments);
/* init */
op = gt_option_parser_new("[options] archive",
"Decompresses condenseq archive.");
/* -seq */
optionseq = gt_option_new_uword("seq",
"only 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",
"only extract (inclusive) range of "
"consecutive sequences identified by "
"their zero based index numbers",
&arguments->seqrange, NULL);
gt_option_parser_add_option(op, optionseqrange);
gt_option_exclude(optionseq, optionseqrange);
/* -range */
optionrange = gt_option_new_range("range",
"only extract (inclusive) range of zero "
"based positions "
"(implies option -output concat)",
&arguments->range, NULL);
gt_option_parser_add_option(op, optionrange);
gt_option_exclude(optionseq, optionrange);
gt_option_exclude(optionseqrange, optionrange);
/* -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);
gt_option_imply(optionrange, optionmode);
/* -sepchar */
option = gt_option_new_string("sepchar",
"specify character to print as SEPARATOR "
"(implies option -output concat",
arguments->sepchar, "|");
gt_option_parser_add_option(op, option);
gt_option_imply(option, optionmode);
arguments->sepchar_opt = gt_option_ref(option);
gt_output_file_info_register_options(arguments->ofi, op, &arguments->outfp);
/* -width */
option = gt_option_new_width(&arguments->width);
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 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_sketch_option_parser_new(void *tool_arguments)
{
GtSketchArguments *arguments = tool_arguments;
GtOptionParser *op;
GtOption *option, *option2;
static const char *formats[] = { "png",
#ifdef CAIRO_HAS_PDF_SURFACE
"pdf",
#endif
#ifdef CAIRO_HAS_SVG_SURFACE
"svg",
#endif
#ifdef CAIRO_HAS_PS_SURFACE
"ps",
#endif
NULL
};
static const char *inputs[] = {
"gff",
"bed",
"gtf",
NULL
};
gt_assert(arguments);
/* init */
op = gt_option_parser_new("[option ...] image_file [GFF3_file ...]",
"Create graphical representation of GFF3 "
"annotation files.");
/* -pipe */
option = gt_option_new_bool("pipe", "use pipe mode (i.e., show all gff3 "
"features on stdout)", &arguments->pipe, false);
gt_option_parser_add_option(op, option);
/* -flattenfiles */
option = gt_option_new_bool("flattenfiles", "do not group tracks by source "
"file name and remove file names from track "
"description", &arguments->flattenfiles, false);
gt_option_parser_add_option(op, option);
/* -seqid */
option = gt_option_new_string("seqid", "sequence region identifier\n"
"default: first one in file",
arguments->seqid, NULL);
gt_option_parser_add_option(op, option);
gt_option_hide_default(option);
/* -start */
option = gt_option_new_uword_min("start", "start position\n"
"default: first region start",
&arguments->start, GT_UNDEF_UWORD, 1);
gt_option_parser_add_option(op, option);
gt_option_hide_default(option);
/* -end */
option2 = gt_option_new_uword("end", "end position\ndefault: last region end",
&arguments->end, GT_UNDEF_UWORD);
gt_option_parser_add_option(op, option2);
/* -start and -end must be given together */
gt_option_imply(option, option2);
gt_option_imply(option2, option);
gt_option_hide_default(option2);
/* -width */
option = gt_option_new_uint_min("width", "target image width (in pixel)",
&arguments->width,
800, 1);
gt_option_parser_add_option(op, option);
/* -style */
option = gt_option_new_string("style", "style file to use",
arguments->stylefile,
gt_str_get(arguments->stylefile));
gt_option_parser_add_option(op, option);
/* -format */
option = gt_option_new_choice("format", "output graphics format\n"
"choose from png"
#ifdef CAIRO_HAS_PDF_SURFACE
"|pdf"
#endif
#ifdef CAIRO_HAS_SVG_SURFACE
"|svg"
#endif
#ifdef CAIRO_HAS_PS_SURFACE
"|ps"
#endif
"",
arguments->format, formats[0], formats);
gt_option_parser_add_option(op, option);
/* -input */
option = gt_option_new_choice("input", "input data format\n"
"choose from gff|bed|gtf",
arguments->input, inputs[0], inputs);
gt_option_parser_add_option(op, option);
/* -addintrons */
option = gt_option_new_bool("addintrons", "add intron features between "
"existing exon features (before drawing)",
&arguments->addintrons, false);
gt_option_parser_add_option(op, option);
/* -unsafe */
option = gt_option_new_bool("unsafe", "enable unsafe mode for style file",
&arguments->unsafe, false);
gt_option_parser_add_option(op, option);
/* -showrecmaps */
option = gt_option_new_bool("showrecmaps",
"show RecMaps after image creation",
&arguments->showrecmaps, false);
gt_option_is_development_option(option);
gt_option_parser_add_option(op, option);
/* -streams */
option = gt_option_new_bool("streams", "use streams to write data to file",
&arguments->use_streams, false);
gt_option_is_development_option(option);
gt_option_parser_add_option(op, option);
/* -v */
option = gt_option_new_verbose(&arguments->verbose);
gt_option_parser_add_option(op, option);
/* -force */
option = gt_option_new_bool(GT_FORCE_OPT_CSTR, "force writing to output file",
&arguments->force, false);
gt_option_parser_add_option(op, option);
gt_option_parser_set_min_args(op, 1);
return op;
}
static GtOptionParser* gt_seed_extend_option_parser_new(void *tool_arguments)
{
GtSeedExtendArguments *arguments = tool_arguments;
GtOptionParser *op;
GtOption *option, *op_gre, *op_xdr, *op_cam, *op_his, *op_dif, *op_pmh,
*op_len, *op_err, *op_xbe, *op_sup, *op_frq, *op_mem, *op_ali, *op_bia;
gt_assert(arguments != NULL);
/* init */
op = gt_option_parser_new("[option ...] encseq_basename [encseq_basename]",
"Calculate local alignments using the seed and "
"extend algorithm.");
/* DIAGBANDSEED OPTIONS */
/* -ii */
option = gt_option_new_string("ii",
"Input index for encseq encoded sequences",
arguments->dbs_indexname,
"");
gt_option_is_mandatory(option);
gt_option_parser_add_option(op, option);
/* -qii */
option = gt_option_new_string("qii",
"Query input index (encseq)",
arguments->dbs_queryname,
"");
gt_option_parser_add_option(op, option);
/* -seedlength */
op_len = gt_option_new_uint_min_max("seedlength",
"Minimum length of a seed",
&arguments->dbs_seedlength,
14UL, 1UL, 32UL);
gt_option_parser_add_option(op, op_len);
/* -diagbandwidth */
option = gt_option_new_uword("diagbandwidth",
"Logarithm of diagonal band width (for filter)",
&arguments->dbs_logdiagbandwidth,
6UL);
gt_option_parser_add_option(op, option);
/* -mincoverage */
option = gt_option_new_uword("mincoverage",
"Minimum coverage in two neighbouring diagonal "
"bands (for filter)",
&arguments->dbs_mincoverage,
35UL);
gt_option_parser_add_option(op, option);
/* -maxfreq */
op_frq = gt_option_new_uword_min("maxfreq",
"Maximum frequency of a k-mer (for filter)",
&arguments->dbs_maxfreq,
GT_UWORD_MAX, 1UL);
gt_option_parser_add_option(op, op_frq);
/* -t */
op_sup = gt_option_new_uword_min("t",
"Suppress k-mers occurring at least t times "
"(for filter)",
&arguments->dbs_suppress,
GT_UWORD_MAX, 2UL);
gt_option_exclude(op_sup, op_frq);
gt_option_is_development_option(op_sup);
gt_option_parser_add_option(op, op_sup);
/* -memlimit */
op_mem = gt_option_new_string("memlimit",
"Maximum memory usage to determine the maximum "
"frequency of a k-mer (for filter)",
arguments->dbs_memlimit_str,
"");
gt_option_parser_add_option(op, op_mem);
/* -debug-kmer */
option = gt_option_new_bool("debug-kmer",
"Output KmerPos lists",
&arguments->dbs_debug_kmer,
false);
gt_option_is_development_option(option);
gt_option_parser_add_option(op, option);
/* -debug-seedpair */
option = gt_option_new_bool("debug-seedpair",
"Output SeedPair lists",
&arguments->dbs_debug_seedpair,
false);
gt_option_is_development_option(option);
gt_option_parser_add_option(op, option);
/* -verify */
option = gt_option_new_bool("verify",
"Check that k-mer seeds occur in the sequences",
&arguments->dbs_verify,
false);
gt_option_is_development_option(option);
gt_option_parser_add_option(op, option);
/* SEED EXTENSION OPTIONS */
/* -extendxdrop */
op_xdr = gt_option_new_uword_min_max("extendxdrop",
"Extend seed to both sides using xdrop "
"algorithm, optional parameter "
"specifies sensitivity",
&arguments->se_extendxdrop,
97UL, 90UL, 100UL);
gt_option_argument_is_optional(op_xdr);
gt_option_parser_add_option(op, op_xdr);
arguments->se_option_xdrop = gt_option_ref(op_xdr);
/* -xdropbelow */
op_xbe = gt_option_new_word("xdropbelow",
"Specify xdrop cutoff score (0 means "
"automatically defined depending on minidentity)",
&arguments->se_xdropbelowscore,
0L);
gt_option_imply(op_xbe, op_xdr);
gt_option_parser_add_option(op, op_xbe);
/* -extendgreedy */
op_gre = gt_option_new_uword_min_max("extendgreedy",
"Extend seed to both sides using greedy "
"algorithm, optional parameter "
"specifies sensitivity",
&arguments->se_extendgreedy,
97UL, 90UL, 100UL);
gt_option_argument_is_optional(op_gre);
gt_option_exclude(op_gre, op_xdr);
gt_option_parser_add_option(op, op_gre);
arguments->se_option_greedy = gt_option_ref(op_gre);
/* -history */
op_his = gt_option_new_uword_min_max("history",
"Size of (mis)match history in range [1"
"..64] (trimming for greedy extension)",
&arguments->se_historysize,
60UL, 1UL, 64UL);
gt_option_imply(op_his, op_gre);
gt_option_parser_add_option(op, op_his);
/* -maxalilendiff */
op_dif = gt_option_new_uword("maxalilendiff",
"Maximum difference of alignment length "
"(trimming for greedy extension)",
&arguments->se_maxalilendiff, 0UL);
gt_option_imply(op_dif, op_gre);
gt_option_is_development_option(op_dif);
gt_option_parser_add_option(op, op_dif);
/* -percmathistory */
op_pmh = gt_option_new_uword_min_max("percmathistory",
"percentage of matches required in "
"history (for greedy extension)",
&arguments->se_perc_match_hist,
0UL, 1UL, 100UL);
gt_option_imply(op_pmh, op_gre);
gt_option_is_development_option(op_pmh);
gt_option_parser_add_option(op, op_pmh);
/* -bias-parameters */
op_bia = gt_option_new_bool("bias-parameters",
"Use -maxalilendiff 30 and let percmathistory "
"depend on minidentiy and DNA base distribution",
&arguments->bias_parameters,
false);
gt_option_imply(op_bia, op_gre);
gt_option_exclude(op_bia, op_pmh);
gt_option_exclude(op_bia, op_dif);
gt_option_is_development_option(op_bia);
gt_option_parser_add_option(op, op_bia);
/* -cam */
op_cam = gt_option_new_string("cam",
gt_cam_extendgreedy_comment(),
arguments->se_char_access_mode,
"");
gt_option_is_development_option(op_cam);
gt_option_parser_add_option(op, op_cam);
/* -l */
op_len = gt_option_new_uword_min("l",
"Minimum alignment length "
"(for seed extension)",
&arguments->se_alignlength,
20UL, 1UL);
gt_option_imply_either_2(op_len, op_xdr, op_gre);
gt_option_parser_add_option(op, op_len);
/* -minidentity */
op_err = gt_option_new_uword_min_max("minidentity",
"Minimum identity of matches "
"(for seed extension)",
&arguments->se_minidentity,
80UL, GT_EXTEND_MIN_IDENTITY_PERCENTAGE,
99UL);
gt_option_imply_either_2(op_err, op_xdr, op_gre);
gt_option_parser_add_option(op, op_err);
/* -a */
op_ali = gt_option_new_uword_min("a",
"show alignments/sequences (optional "
"argument is number of columns per line)",
&arguments->se_alignmentwidth,
70, 20);
gt_option_argument_is_optional(op_ali);
gt_option_parser_add_option(op, op_ali);
arguments->se_option_withali = gt_option_ref(op_ali);
/* -mirror */
option = gt_option_new_bool("mirror",
"Add reverse complement reads",
&arguments->mirror,
false);
gt_option_parser_add_option(op, option);
/* -overlappingseeds */
option = gt_option_new_bool("overlappingseeds",
"Allow overlapping SeedPairs",
&arguments->overlappingseeds,
false);
gt_option_is_development_option(option);
gt_option_parser_add_option(op, option);
/* -benchmark */
option = gt_option_new_bool("benchmark",
"Measure total running time and be silent",
&arguments->benchmark,
false);
gt_option_is_development_option(option);
gt_option_parser_add_option(op, option);
/* -seed-display */
option = gt_option_new_bool("seed-display",
"Display seeds in #-line",
&arguments->seed_display,
false);
gt_option_is_development_option(option);
gt_option_parser_add_option(op, option);
/* -v */
option = gt_option_new_verbose(&arguments->verbose);
gt_option_parser_add_option(op, option);
return op;
}
static GtOptionParser* gt_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 */
static const char *overlaps[] = {
"best", /* default */
"no",
"all",
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_ulong_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_ulong_min_max("mintirlen",
"specify minimum length for "
"each TIR",
&arguments->min_TIR_length,
27UL, 1UL, GT_UNDEF_ULONG);
gt_option_parser_add_option(op, optionminlentir);
/* -maxlentir */
optionmaxlentir = gt_option_new_ulong_min_max("maxtirlen",
"specify maximum length for "
"each TIR",
&arguments->max_TIR_length,
1000UL, 1UL, GT_UNDEF_ULONG);
gt_option_parser_add_option(op, optionmaxlentir);
/* -mindisttir */
optionmindisttir = gt_option_new_ulong_min_max("mintirdist",
"specify minimum distance of "
"TIRs",
&arguments->min_TIR_distance,
100UL, 1UL, GT_UNDEF_ULONG);
gt_option_parser_add_option(op, optionmindisttir);
/* -maxdisttir */
optionmaxdisttir = gt_option_new_ulong_min_max("maxtirdist",
"specify maximum distance of "
"TIRs",
&arguments->max_TIR_distance,
10000UL, 1UL, GT_UNDEF_ULONG);
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_ulong_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_ulong_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_ulong_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);
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;
}
static GtOptionParser* gt_eval_option_parser_new(void *tool_arguments)
{
EvalArguments *arguments = tool_arguments;
GtOptionParser *op;
GtOption *option, *ltroption, *ltrdeltaoption;
gt_assert(arguments);
op = gt_option_parser_new("reference_file prediction_file ",
"Compare annotation files and show "
"accuracy measures (prediction vs. reference).");
/* -exondiff */
option = gt_option_new_bool("exondiff", "show a diff for the exons",
&arguments->exondiff, false);
gt_option_is_development_option(option);
gt_option_parser_add_option(op, option);
/* -exondiffcollapsed */
option = gt_option_new_bool("exondiffcollapsed", "show a diff for the "
"collapsed exons", &arguments->exondiffcollapsed,
false);
gt_option_is_development_option(option);
gt_option_parser_add_option(op, option);
/* -nuc */
option = gt_option_new_bool("nuc",
"evaluate nucleotide level (memory consumption "
"is proportional to the input file sizes)",
&arguments->nuceval, true);
gt_option_parser_add_option(op, option);
/* -ltr */
ltroption = gt_option_new_bool("ltr", "evaluate a LTR retrotransposon "
"prediction instead of a gene prediction\n"
"(all LTR_retrotransposon elements are "
"considered to have an undetermined strand)",
&arguments->evalLTR, false);
gt_option_parser_add_option(op, ltroption);
/* -ltrdelta */
ltrdeltaoption = gt_option_new_uword("ltrdelta", "set allowed delta for LTR "
"borders to be considered equal",
&arguments->LTRdelta, 20);
gt_option_parser_add_option(op, ltrdeltaoption);
/* -v */
option = gt_option_new_verbose(&arguments->verbose);
gt_option_parser_add_option(op, option);
/* output file options */
gt_output_file_info_register_options(arguments->ofi, op, &arguments->outfp);
/* option implications */
gt_option_imply(ltrdeltaoption, ltroption);
/* set comment function */
gt_option_parser_set_comment_func(op, gt_gtdata_show_help, NULL);
/* set minimum and maximum number of arguments */
gt_option_parser_set_min_max_args(op, 2, 2);
return op;
}