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_ulong("width", "set output width for showing of "
"sequences (0 disables formatting)",
&arguments->width, 0);
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_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* 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_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);
/* 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;
}
