static GtOptionParser* gt_splicesiteinfo_option_parser_new(void *tool_arguments)
{
SpliceSiteInfoArguments *arguments = tool_arguments;
GtOptionParser *op;
GtOption *option;
gt_assert(arguments);
op = gt_option_parser_new("[option ...] [GFF3_file ...]", "Show information "
"about splice sites given in GFF3 files.");
/* -seqfile, -matchdesc, -usedesc and -regionmapping */
gt_seqid2file_register_options(op, arguments->s2fi);
/* -addintrons */
option = gt_option_new_bool("addintrons", "add intron features between "
"existing exon features\n(before computing the "
"information to be shown)",
&arguments->addintrons, false);
gt_option_parser_add_option(op, option);
/* output file options */
gt_outputfile_register_options(op, &arguments->outfp, arguments->ofi);
gt_option_parser_set_comment_func(op, gt_gtdata_show_help, NULL);
return op;
}
static GtOptionParser* gt_seqtransform_option_parser_new(void *tool_arguments)
{
SeqtransformArguments *arguments = tool_arguments;
GtOptionParser *op;
GtOption *o;
gt_assert(arguments);
op = gt_option_parser_new("[option ...] [sequence_file ...]",
"Perform simple transformations on the given "
"sequence file(s).");
/* -addstopaminos */
o = gt_option_new_bool("addstopaminos", "append stop amino acids ('"
GT_STOP_AMINO_CSTR"') to given protein sequences, if "
"not already present", &arguments->addstopaminos,
false);
gt_option_parser_add_option(op, o);
/* -width */
o = gt_option_new_width(&arguments->width);
gt_option_parser_add_option(op, o);
gt_outputfile_register_options(op, &arguments->outfp, arguments->ofi);
return op;
}
static GtOptionParser* gt_cds_option_parser_new(void *tool_arguments)
{
CDSArguments *arguments = tool_arguments;
GtOptionParser *op;
GtOption *option;
gt_assert(arguments);
op = gt_option_parser_new("[option ...] [GFF3_file]",
"Add CDS (coding sequence) features to exon "
"features given in GFF3 file.");
/* -minorflen */
option = gt_option_new_uint_min("minorflen", "set the minimum length an open "
"reading frame (ORF) must have to be added "
"as a CDS feature (measured in amino acids)",
&arguments->minorflen, 64, 1);
gt_option_parser_add_option(op, option);
/* -startcodon */
option = gt_option_new_bool("startcodon", "require than an ORF must begin "
"with a start codon", &arguments->start_codon,
false);
gt_option_parser_add_option(op, option);
/* -finalstopcodon */
option = gt_option_new_bool("finalstopcodon", "require that the final ORF "
"must end with a stop codon",
&arguments->final_stop_codon, false);
gt_option_parser_add_option(op, option);
/* -genericstartcodons */
option = gt_option_new_bool("genericstartcodons", "use generic start codons",
&arguments->generic_start_codons, false);
gt_option_is_development_option(option);
gt_option_parser_add_option(op, option);
/* -seqfile, -matchdesc, -usedesc and -regionmapping */
gt_seqid2file_register_options(op, arguments->s2fi);
/* -v */
option = gt_option_new_verbose(&arguments->verbose);
gt_option_parser_add_option(op, option);
/* output file options */
gt_outputfile_register_options(op, &arguments->outfp, arguments->ofi);
gt_option_parser_set_comment_func(op, gt_gtdata_show_help, NULL);
gt_option_parser_set_max_args(op, 1);
return op;
}
static GtOptionParser* gt_mergefeat_option_parser_new(void *tool_arguments)
{
InterFeatArguments *arguments = tool_arguments;
GtOptionParser *op;
gt_assert(arguments);
/* init */
op = gt_option_parser_new("[option ...] [GFF3_file ...]", "Merge adjacent "
"features of the same type in given GFF3_file(s).");
/* output file options */
gt_outputfile_register_options(op, &arguments->outfp, arguments->ofi);
return op;
}
static GtOPrval parse_options(int *parsed_args, ChseqidsArguments *arguments,
int argc, const char **argv, GtError *err)
{
GtOptionParser *op;
GtOutputFileInfo *ofi;
GtOption *option;
GtOPrval oprval;
gt_error_check(err);
/* init */
op = gt_option_parser_new("[option ...] mapping_file [GFF3_file]",
"Change sequence ids by the mapping given in "
"mapping_file.");
ofi = gt_outputfileinfo_new();
/* -sort */
option = gt_option_new_bool("sort",
"sort the GFF3 features after changing the "
"sequence ids\n(memory consumption is "
"proportional to the input file size)",
&arguments->sort, false);
gt_option_parser_add_option(op, option);
/* -v */
option = gt_option_new_verbose(&arguments->verbose);
gt_option_parser_add_option(op, option);
/* output file options */
gt_outputfile_register_options(op, &arguments->outfp, ofi);
/* parse options */
gt_option_parser_set_comment_func(op, gt_gtdata_show_help, NULL);
gt_option_parser_set_min_max_args(op, 1, 2);
oprval = gt_option_parser_parse(op, parsed_args, argc, argv, gt_versionfunc,
err);
/* free */
gt_outputfileinfo_delete(ofi);
gt_option_parser_delete(op);
return oprval;
}
static GtOptionParser* gt_seqfilter_option_parser_new(void *tool_arguments)
{
SeqFilterArguments *arguments = tool_arguments;
GtOption *option;
GtOptionParser *op;
gt_assert(arguments);
op = gt_option_parser_new("[option ...] [sequence_file ...]",
"Filter the given sequence_file(s) and show the "
"results on stdout.");
/* -minlength */
option = gt_option_new_ulong("minlength",
"set minimum length a sequence must "
"have to pass the filter", &arguments->minlength,
GT_UNDEF_ULONG);
gt_option_parser_add_option(op, option);
/* -maxlength */
option = gt_option_new_ulong("maxlength", "set maximum length a sequence can "
"have to pass the filter", &arguments->maxlength,
GT_UNDEF_ULONG);
gt_option_parser_add_option(op, option);
/* -maxseqnum */
option = gt_option_new_ulong("maxseqnum", "set the maximum number of "
"sequences which can pass the filter",
&arguments->maxseqnum, GT_UNDEF_ULONG);
gt_option_parser_add_option(op, option);
/* -width */
option = gt_option_new_width(&arguments->width);
gt_option_parser_add_option(op, option);
gt_outputfile_register_options(op, &arguments->outfp, arguments->ofi);
return op;
}
static GtOptionParser* gt_seqmutate_option_parser_new(void *tool_arguments)
{
MutateArguments *arguments = tool_arguments;
GtOptionParser *op;
GtOption *o;
gt_assert(arguments);
op = gt_option_parser_new("[option ...] [sequence_file ...]",
"Mutate the sequences of the given sequence "
"file(s).");
/* -rate */
o = gt_option_new_uint_max("rate", "set the mutation rate", &arguments->rate,
1, 100);
gt_option_parser_add_option(op, o);
/* -width */
o = gt_option_new_width(&arguments->width);
gt_option_parser_add_option(op, o);
gt_outputfile_register_options(op, &arguments->outfp, arguments->ofi);
gt_option_parser_set_comment_func(op, gt_gtdata_show_help, NULL);
return op;
}
static GtOptionParser* gt_extractseq_option_parser_new(void *tool_arguments)
{
ExtractSeqArguments *arguments = tool_arguments;
GtOptionParser *op;
GtOption *frompos_option, *topos_option, *match_option, *width_option,
*fastakeyfile_option;
gt_assert(arguments);
/* init */
op = gt_option_parser_new("[option ...] [sequence_file(s)] | fastaindex",
"Extract sequences from given sequence file(s) or "
"fastaindex.");
/* -frompos */
frompos_option = gt_option_new_ulong_min(FROMPOS_OPTION_STR,
"extract sequence from this position\n"
"counting from 1 on",
&arguments->frompos, 0, 1UL);
gt_option_parser_add_option(op, frompos_option);
/* -topos */
topos_option = gt_option_new_ulong_min(TOPOS_OPTION_STR,
"extract sequence up to this position\n"
"counting from 1 on",
&arguments->topos, 0, 1UL);
gt_option_parser_add_option(op, topos_option);
/* -match */
match_option = gt_option_new_string("match", "extract all sequences whose "
"description matches the given pattern.\n"
"The given pattern must be a valid extended "
"regular expression.", arguments->pattern,
NULL);
gt_option_parser_add_option(op, match_option);
/* -keys */
fastakeyfile_option = gt_option_new_filename("keys",
"extract substrings for keys "
"in specified file",
arguments->fastakeyfile);
gt_option_parser_add_option(op, fastakeyfile_option);
/* -width */
width_option = gt_option_new_width(&arguments->width);
gt_option_parser_add_option(op, width_option);
/* output file options */
gt_outputfile_register_options(op, &arguments->outfp, arguments->ofi);
/* option implications */
gt_option_imply(frompos_option, topos_option);
gt_option_imply(topos_option, frompos_option);
/* option exclusions */
gt_option_exclude(frompos_option, match_option);
gt_option_exclude(topos_option, match_option);
gt_option_exclude(frompos_option, fastakeyfile_option);
gt_option_exclude(match_option, fastakeyfile_option);
gt_option_parser_set_comment_func(op, gt_gtdata_show_help, NULL);
return op;
}
static GtOptionParser* gt_ltrdigest_option_parser_new(void *tool_arguments)
{
GtLTRdigestOptions *arguments = tool_arguments;
GtOptionParser *op;
GtOption *o, *ot, *oto;
#ifdef HAVE_HMMER
GtOption *oh, *oc, *oeval;
static const char *cutoffs[] = {"NONE", "GA", "TC", NULL};
#endif
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 indexname",
"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_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_opts.ppt_len,
&pptlen_defaults);
gt_option_parser_add_option(op, o);
o = gt_option_new_range("uboxlen",
"required U-box length range",
&arguments->ppt_opts.ubox_len,
&uboxlen_defaults);
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_opts.radius,
30U);
gt_option_parser_add_option(op, o);
o = gt_option_new_probability("pptrprob",
"purine emission probability inside PPT",
&arguments->ppt_opts.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_opts.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->ppt_opts.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->ppt_opts.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->ppt_opts.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->ppt_opts.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->ppt_opts.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->pbs_opts.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->pbs_opts.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->pbs_opts.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->pbs_opts.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_opts.radius,
30U);
gt_option_parser_add_option(op, o);
gt_option_imply(o, ot);
/* Protein domain search options */
#ifdef HAVE_HMMER
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->pdom_opts.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->pdom_opts.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->pdom_opts.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->pdom_opts.write_aaseqs,
false);
gt_option_parser_add_option(op, o);
gt_option_imply(o, oh);
gt_option_imply(o, oto);
o = gt_option_new_uint("maxgaplen",
"maximal allowed gap size between fragments (in amino "
"acids) when chaining pHMM hits for a protein domain",
&arguments->pdom_opts.chain_max_gap_length,
50);
gt_option_parser_add_option(op, o);
gt_option_is_extended_option(o);
gt_option_imply(o, oh);
o = gt_option_new_ulong("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);
#endif
/* Extended PBS options */
o = gt_option_new_int("pbsmatchscore",
"match score for PBS/tRNA alignments",
&arguments->pbs_opts.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->pbs_opts.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->pbs_opts.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->pbs_opts.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_outputfile_register_options(op, &arguments->outfp, arguments->ofi);
gt_option_parser_set_min_max_args(op, 2U, 2U);
return op;
}
