RetroSearch Browse

Home - News ( United States | United Kingdom | Italy | Germany ) - Football scores

Showing content from http://www.ncbi.nlm.nih.gov/IEB/ToolBox/CPP_DOC/doxyhtml/hitfilter__app_8cpp_source.html below:

NCBI C++ ToolKit: src/app/hfilter/hitfilter_app.cpp Source File

58 const string

g_m8(

"m8"

), g_AsnTxt(

"asntxt"

), g_AsnBin(

"asnbin"

);

60 const string

kMode_Pairwise (

"pairwise"

);

61 const string

kMode_Multiple (

"multiple"

);

66 const string

kBoth(

"strict"

);

67 const string kQuery

(

"query"

);

76 "HitFilter v.2.0.2"

);

78

argdescr->AddDefaultKey(

"mode"

79 "Specify whether the hits should be resolved in pairs " 80 "or as a single set."

84

argdescr->AddDefaultKey(

"min_idty"

85 "Minimal input hit identity"

88

argdescr->AddDefaultKey(

"min_len"

89 "Minimal input hit length"

92

argdescr->AddDefaultKey(

"retain_overlap"

93 "Min overlap to retain in kilobases (0=OFF)"

96

argdescr->AddDefaultKey(

"fmt_in"

"Input format"

99

argdescr->AddOptionalKey(

"file_in"

"Input file (stdin otherwise)"

103

argdescr->AddFlag(

"sas"

"Assume seq-align-set as the top-level structure " 104 "for the input ASN hits"

true

);

106

argdescr->AddDefaultKey(

"merge"

107 "Merge abutting alignments unless the merged " 108 "alignment overlap length ratio is greater " 109 "than this parameter. Any negative value will " 110 "turn merging off."

114

argdescr->AddOptionalKey(

"constraints"

115 "Binary ASN file with constraining alignments"

119

argdescr->AddOptionalKey(

"file_out"

"Output file (stdout otherwise)"

123

argdescr->AddOptionalKey(

"m"

124 "Text description/comment to add to the output"

127

argdescr->AddDefaultKey(

"fmt_out"

"Output format"

130

argdescr->AddDefaultKey(

"hits_per_chunk"

131 "Input is split into chunks with the number of hits " 132 "per chunk limited by this parameter."

136

argdescr->AddDefaultKey(

"coord_margin"

137 "Larger values of this argument will result in less " 138 "RAM used but longer running times."

142

argdescr->AddOptionalKey(

"ids"

"Table to rename sequence IDs."

145

argdescr->AddDefaultKey(

"ut"

"uniqueness_type"

146 "uniqueness type (strict, query, or subject)"

150

unique_type->

Allow

(

"strict"

)->

Allow

(

"query"

)->

Allow

(

"subject"

);

151

argdescr->SetConstraint(

"ut"

, unique_type);

153

argdescr->AddFlag(

"keep_strands"

154 "Keep plus-plus strands" 157

argdescr->AddFlag(

"no_output_constraint"

158 "Do not output constraints" 162

argdescr->SetConstraint(

"fmt_in"

, constrain_format);

163

argdescr->SetConstraint(

"fmt_out"

, constrain_format);

166

argdescr->SetConstraint(

"min_len"

, constrain_minlen);

169

argdescr->SetConstraint(

"min_idty"

, constrain_minidty);

172

argdescr->SetConstraint(

"merge"

, constrain_merge);

175

constrain_mode->

Allow

(kMode_Pairwise)->

Allow

(kMode_Multiple);

176

argdescr->SetConstraint(

"mode"

, constrain_mode);

187 string

ctgid, accver;

189 if

(ctgid.size() == 0) {

193 if

(accver.size() == 0) {

196 m_IDs

[ctgid] = accver;

201

build_ids.

m_id

[0] = build_ids.

m_id

[1] = ctgid;

206

bi.

m_id

[1] = ctgid;

216 const string

fmt_in = args[

"fmt_in"

].AsString();

217 const string

fmt_out = args[

"fmt_out"

].AsString();

218 const THit::TCoord

min_len = args[

"min_len"

].AsInteger();

219 const double

min_idty = args[

"min_idty"

].AsDouble();

221 CNcbiIstream

& istr = args[

"file_in"

]? args[

"file_in"

].AsInputFile(): cin;

225 if

(fmt_in == g_m8) {

227 static string

firstline;

230 if

(one_pair && firstline.size()) {

233 if

(hit->GetIdentity() >= min_idty && hit->GetLength() >= min_len) {

234

phitrefs->push_back(hit);

235

id_query = hit->GetQueryId();

236

id_subj = hit->GetSubjId();

253 if

(id_query.

IsNull

()) {

254

id_query = hit->GetQueryId();

255

id_subj = hit->GetSubjId();

257 else if

(

false

== id_query -> Match(*(hit->GetQueryId()))

258

false

== id_subj -> Match(*(hit->GetSubjId())) )

260 if

(phitrefs->size()) {

265

id_query = hit->GetQueryId();

266

id_subj = hit->GetSubjId();

271 if

(hit->GetIdentity() >= min_idty && hit->GetLength() >= min_len) {

272

phitrefs->push_back(hit);

279 const bool

parse_aln = fmt_out != g_m8;

283

unique_ptr<CObjectIStream>

(in_ptr);

285 const bool

assume_sas (args[

"sas"

]);

287 while

->EndOfData()) {

297

phitrefs, parse_aln, min_len, min_idty);

310 if

(one_pair && phitrefs->size()) {

315 static

TStringSet idtags;

317 const string

strid_query (phitrefs->front()->GetId(0)->GetSeqIdString(

true

));

318 const string

strid_subj (phitrefs->front()->GetId(1)->GetSeqIdString(

true

));

319 const string tag

(strid_subj +

"$_#_&"

+ strid_query);

320 if

(idtags.end() != idtags.find(

tag

)) {

322 "In pairwise mode input hits must be collated " 323 "by query and subject."

);

326

idtags.insert(

tag

);

336 const double

& min_idty)

const 342 if

(

.GetTo() -

.GetFrom() >= min_len) {

345 if

(hit->GetIdentity() >= min_idty) {

346 if

(hit->GetQueryStrand() ==

false

) {

349

phitrefs->push_back(hit);

358 const string

fmt = args[

"fmt_out"

].AsString();

360 CNcbiOstream

& ostr = args[

"file_out"

]? args[

"file_out"

].AsOutputFile(): cout;

362 string

comment (args[

"m"

]? args[

"m"

].AsString():

""

);

366 if

(comment.size() > 0) {

367

ostr <<

"# "

<< comment << endl;

375

"\tNumGapOpenings" 385 const THit

& hit = **ii;

394 const bool

fmt_txt (fmt == g_AsnTxt);

396 const THit

& h = **ii;

398 bool

no_output_constraint = args[

"no_output_constraint"

].HasValue();

399 if

(no_output_constraint && h.

GetScore

() >= kBigDbl) {

414 bool

is_gap =

false

;

416 for

(

int i

= 0;

< ds->

GetDim

();

++) {

431 bool

keep_strands = args[

"keep_strands"

].HasValue();

436

vector< CRef< CSeq_id > > &ids = ds->

SetIds

();

437 for

(

Uint1

where = 0; where < 2; ++where) {

440 id

->Assign(*h.

GetId

(where));

449

score->SetId().SetStr(

"reciprocity"

);

451 if

(h.

GetScore

() > kBigDbl || args[

"ut"

].AsString() == kBoth)

456

}

else if

(args[

"ut"

].AsString() ==

kQuery

) {

463

cerr <<

"Error adding reciprocity"

<< endl;

466

seq_align->

SetScore

().push_back(score);

469

seq_align->

SetSegs

().SetDenseg(*ds);

471

align_list.push_back(seq_align);

474 if

(comment.size() > 0) {

487

cerr <<

"Error writing output file"

<< endl;

496 return

lhs->GetScore() > rhs->GetScore();

504 const bool

mode_multiple ( args[

"mode"

].AsString() == kMode_Multiple );

505 const string

fmt_in ( args[

"fmt_in"

].AsString() );

506 const string

fmt_out ( args[

"fmt_out"

].AsString() );

507 const double

maxlenfr (args[

"merge"

].AsDouble());

509 if

((fmt_out == g_AsnTxt || fmt_out == g_AsnBin) &&

510

(fmt_in != g_AsnTxt && fmt_in != g_AsnBin))

514 "For ASN output, input must also be in ASN"

);

517 if

( mode_multiple ==

false

&& (args[

"ids"

] || args[

"constraints"

]

518

|| fmt_in == g_AsnTxt || fmt_in == g_AsnBin ))

522 "Invalid parameter combination - " 523 "some options are not yet supported in pairwise mode."

);

553 const THit::TCoord

min_len (args[

"min_len"

].AsInteger());

554 const double

min_idty (args[

"min_idty"

].AsDouble());

555 const size_t

margin (args[

"coord_margin"

].AsInteger());

556 const THit::TCoord

retain_overlap (1024 * args[

"retain_overlap"

].AsInteger());

559 if

(args[

"ut"

].AsString() ==

"query"

) {

561

}

else if

(args[

"ut"

].AsString() ==

"subject"

) {

571

hits.begin(), hits.end(),

580 copy

(hits.begin(), hits.end(), back_inserter(

all

));

581 copy

(hits_new.begin(), hits_new.end(), back_inserter(

all

));

586

cerr <<

"Error running x_DoPairwise"

<< endl;

598 const string

fmt_in = args[

"fmt_in"

].AsString();

599 const string

fmt_out = args[

"fmt_out"

].AsString();

600 const THit::TCoord

min_len = args[

"min_len"

].AsInteger();

601 const double

min_idty = args[

"min_idty"

].AsDouble();

602 const THit::TCoord

retain_overlap = 1024 * args[

"retain_overlap"

].AsInteger();

603 const size_t

margin (args[

"coord_margin"

].AsInteger());

606 if

(args[

"ut"

].AsString() ==

"query"

) {

608

}

else if

(args[

"ut"

].AsString() ==

"subject"

) {

617 if

(args[

"constraints"

]) {

622 copy

(restraint.begin(), restraint.end(), back_inserter(

all

));

626 const size_t M

= args[

"hits_per_chunk"

].AsInteger();

627 const size_t

dim =

all

.size();

628 size_t

m =

min

(dim,

);

630 const

THitRefs::iterator ii_beg =

all

.begin(), ii_end =

all

.end();

631

THitRefs::iterator ii_hi = ii_beg, ii = ii_beg;

634 while

(ii < ii_end) {

636

THitRefs::iterator ii_dst = ii + m;

637 if

(ii_dst > ii_end) {

642 copy

(ii, ii_dst, ii_hi);

643

ii_hi += ii_dst - ii;

658

THitRefs::iterator ii_hi0 = ii_hi;

660

THitRefs::iterator jj = hits_new.begin(), jje = hits_new.end();

661 for

(;jj != jje && ii_hi != ii_hi0; *ii_hi++ = *jj++);

663 LOG_POST

(

"Warning: space from eliminated alignments " 664 "not enough for all splits."

);

669

cerr <<

"Error in x_DoMultiple"

<< endl;

672 all

.erase(ii_hi, ii_end);

696 for

(

Uint1

where = 0; where < 2; ++where) {

716 id

->Assign(*(hit->GetId(where)));

719 const string

ctgid =

string

(

"lcl|"

) + im->second.m_id[where];

720 id

.Reset(

new CSeq_id

(ctgid));

722

hit->SetId(where,

id

);

725 if

(hit->GetQueryStrand() ==

false

) {

729

hit->SetScore(kBigDbl);

733 if

(hit->GetLength() > maxlen) {

734

maxlen = hit->GetLength();

737 float

score_factor = 0.25 / maxlen;

741

h->SetScore(h->GetScore() * (1 + score_factor * h->GetLength()));

743 if

(args[

"no_output_constraint"

HasValue

()) {

761 int main

(

int

argc,

const char

* argv[])

User-defined methods of the data storage class.

void remove_if(Container &c, Predicate *__pred)

bool GetQueryStrand(void) const

TCoord GetQueryStart(void) const

static string s_RunLengthDecode(const string &in)

TCoord GetSubjStart(void) const

bool GetSubjStrand(void) const

const TId & GetId(Uint1 where) const

const TTranscript & GetTranscript(void) const

list< CRef< objects::CSeq_align > > TSeqAlignList

void x_LoadIDs(CNcbiIstream &istr)

void x_LoadConstraints(CNcbiIstream &istr, THitRefs &all)

void x_DoPairwise(THitRefs *pall)

void x_ReadInputHits(THitRefs *phitrefs, bool one_pair=false)

virtual void Exit()

Cleanup on application exit.

vector< THitRef > THitRefs

void x_IterateSeqAlignList(const TSeqAlignList &sa_list, THitRefs *phitrefs, bool parse_aln, const THit::TCoord &min_len, const double &min_idty) const

void x_DumpOutput(const THitRefs &hitrefs)

void x_DoMultiple(THitRefs *pall)

virtual int Run()

Run the application.

virtual void Init()

Initialize the application.

float GetScore(void) const

void FromTranscript(TSeqPos query_start, ENa_strand query_strand, TSeqPos subj_start, ENa_strand subj_strand, const string &transcript)

Initialize from pairwise alignment transcript (a string representation produced by CNWAligner)

static TRegisterLoaderInfo RegisterInObjectManager(CObjectManager &om, CReader *reader=0, CObjectManager::EIsDefault is_default=CObjectManager::eDefault, CObjectManager::TPriority priority=CObjectManager::kPriority_NotSet)

void AddComment(const string &comment)

container_type::const_iterator const_iterator

container_type::iterator iterator

const_iterator end() const

const_iterator find(const key_type &key) const

static void s_RunGreedy(typename THitRefs::iterator hri_beg, typename THitRefs::iterator hri_end, THitRefs *phits_new, TCoord min_hit_len=100, double min_hit_idty=.9, TCoord margin=1, TCoord retain_overlap=0, EUnique_type unique_type=e_Strict)

EUnique_type

Multiple-sequences greedy alignment uniquification algorithm.

static void s_MergeAbutting(typename THitRefs::iterator hri_beg, typename THitRefs::iterator hri_end, const double &maxlenfr, THitRefs *pout)

void HideStdArgs(THideStdArgs hide_mask)

Set the hide mask for the Hide Std Flags.

virtual const CArgs & GetArgs(void) const

Get parsed command line arguments.

int AppMain(int argc, const char *const *argv, const char *const *envp=0, EAppDiagStream diag=eDS_Default, const char *conf=NcbiEmptyCStr, const string &name=NcbiEmptyString)

Main function (entry point) for the NCBI application.

virtual void SetupArgDescriptions(CArgDescriptions *arg_desc)

Setup the command line argument descriptions.

#define ITERATE(Type, Var, Cont)

ITERATE macro to sequence through container elements.

#define NON_CONST_ITERATE(Type, Var, Cont)

Non constant version of ITERATE macro.

const CNcbiArguments & GetArguments(void) const

Get the application's cached unprocessed command-line arguments.

@ fHideLogfile

Hide log file description.

@ fHideConffile

Hide configuration file description.

@ fHideVersion

Hide version description.

CArgAllow_Strings * Allow(const string &value)

Add allowed string values.

@ fBinary

Open as binary file; for eInputFile, eOutputFile, eIOFile.

@ eInputFile

Name of file (must exist and be readable)

@ eDouble

Convertible into a floating point number (double)

@ eString

An arbitrary string.

@ eOutputFile

Name of file (must be writable)

@ eInteger

Convertible into an integer number (int or Int8)

#define LOG_POST(message)

This macro is deprecated and it's strongly recomended to move in all projects (except tests) to macro...

#define NCBI_THROW(exception_class, err_code, message)

Generic macro to throw an exception, given the exception class, error code and message string.

#define MSerial_AsnBinary

#define MSerial_AsnText

I/O stream manipulators â.

@ eSerial_AsnText

ASN.1 text.

@ eSerial_AsnBinary

ASN.1 binary.

const string AsFastaString(void) const

static CObjectIStream * Open(ESerialDataFormat format, CNcbiIstream &inStream, bool deleteInStream)

Create serial object reader and attach it to an input stream.

TGi GetGiForAccession(const string &acc, CScope &scope, EGetIdType flags=0)

Given an accession string retrieve the GI id.

string GetAccessionForGi(TGi gi, CScope &scope, EAccessionVersion use_version=eWithAccessionVersion, EGetIdType flags=0)

Retrieve the accession for a given GI.

static CRef< CObjectManager > GetInstance(void)

Return the existing object manager or create one.

void AddDefaults(TPriority pri=kPriority_Default)

Add default data loaders from object manager.

bool IsNull(void) const THROWS_NONE

Check if pointer is null â same effect as Empty().

uint8_t Uint1

1-byte (8-bit) unsigned integer

#define END_NCBI_SCOPE

End previously defined NCBI scope.

#define BEGIN_NCBI_SCOPE

Define ncbi namespace.

IO_PREFIX::ostream CNcbiOstream

Portable alias for ostream.

IO_PREFIX::istream CNcbiIstream

Portable alias for istream.

static string TruncateSpaces(const string &str, ETrunc where=eTrunc_Both)

Truncate whitespace in a string.

TScore & SetScore(void)

Assign a value to Score data member.

const TStarts & GetStarts(void) const

Get the Starts member data.

void SetSegs(TSegs &value)

Assign a value to Segs data member.

void ResetStrands(void)

Reset Strands data member.

void SetType(TType value)

Assign a value to Type data member.

TDim GetDim(void) const

Get the Dim member data.

TNumseg GetNumseg(void) const

Get the Numseg member data.

TIds & SetIds(void)

Assign a value to Ids data member.

const TDisc & GetDisc(void) const

Get the variant data.

const Tdata & Get(void) const

Get the member data.

const TSegs & GetSegs(void) const

Get the Segs member data.

@ eType_partial

mapping pieces together

ENa_strand

strand of nucleic acid

TGi GetGi(void) const

Get the variant data.

bool IsGi(void) const

Check if variant Gi is selected.

void SetData(TData &value)

Assign a value to Data data member.

list< CRef< CSeq_align > > TAlign

const TAlign & GetAlign(void) const

Get the variant data.

const TData & GetData(void) const

Get the Data member data.

bool s_PHitRefScore(const CAppHitFilter::THitRef &lhs, const CAppHitFilter::THitRef &rhs)

int main(int argc, const char *argv[])

where boath are integers</td > n< td ></td > n</tr > n< tr > n< td > tse</td > n< td > optional</td > n< td > String</td > n< td class=\"description\"> TSE option controls what blob is smart and slim</td> n<td> orig</td> n</tr> n<tr> n<td> last_modified</td> n<td> optional</td> n<td> Integer</td> n<td class=\"description\"> The blob last modification If provided then the exact match will be requested with n the Cassandra storage corresponding field value</td> n<td> Positive integer Not provided means that the most recent match will be selected</td> n<td></td> n</tr> n<tr> n<td> use_cache</td> n<td> optional</td> n<td> String</td> n<td class=\"description\"> The option controls if the Cassandra LMDB cache and or database should be used It n affects the seq id resolution step and the blob properties lookup step The following n options are BIOSEQ_INFO and BLOB_PROP at all

constexpr auto sort(_Init &&init)

#define GetProgramName

Avoid name clash with the NCBI C Toolkit.

std::istream & in(std::istream &in_, double &x_)

double r(size_t dimension_, const Int4 *score_, const double *prob_, double theta_)

void copy(Njn::Matrix< S > *matrix_, const Njn::Matrix< T > &matrix0_)

CRef< objects::CObjectManager > om

RetroSearch is an open source project built by @garambo | Open a GitHub Issue

Search and Browse the WWW like it's 1997 | Search results from DuckDuckGo

HTML: 3.2 | Encoding: UTF-8 | Version: 0.7.4