RetroSearch Browse

->m_msg_skipped <<

"</skipped>\n"

;

129 if

(m_GenBankCompLineCount) {

130 out

<< m_GenBankCompLineCount <<

" lines with valid component accessions"

;

133 out

"No valid component accessions found"

;

136 if

(e_count==1)

".\n1 error"

;

137 else out

".\n"

<< e_count <<

" errors"

;

138 if

(

->m_msg_skipped)

", "

->m_msg_skipped <<

" not printed"

;

143 out

"; no invalid component accessions.\n"

;

152

map_it = m_TaxidSpeciesMap.find(taxid);

155 if

(map_it == m_TaxidSpeciesMap.end()) {

156

species_id = x_GetTaxonSpecies(taxid);

157

m_TaxidSpeciesMap.insert(

160

species_id = map_it->second;

171 bool

is_species =

true

;

185 id

, is_species, is_uncultured, blast_name

187 if

(org_ref ==

null

) {

192 if

(

id

==taxid) blast_name0=blast_name;

197

species_id = prev_id;

202 if

(blast_name0.size()) {

209 string

(

" - "

) + blast_name0 +

210 " is above species level"

);

230

(res.first)->second.push_back(line_info);

231

m_TaxidComponentTotal++;

236 if

(m_TaxidMap.size() == 0)

return true

;

239 float

agp_taxid_percent = 0;

241 float

max_percent = 0;

242 bool

taxid_found=

false

;

246

agp_taxid_percent = float(it->second.size())/float(m_TaxidComponentTotal);

247 if

(agp_taxid_percent > max_percent) {

248

max_percent = agp_taxid_percent;

249

agp_taxid = it->first;

250 if

(agp_taxid_percent>=.8) {

257 if

(use_xml)

" <taxid>"

<< agp_taxid <<

"</taxid>\n"

;

260 out

" <cannot_determine_taxid/>\n"

;

263

cerr <<

"\nUnable to determine a Taxid for the AGP"

;

265

cerr <<

":\nless than 80% of components have one common taxid="

<<agp_taxid<<

""

;

273 if

(!use_xml)

"The AGP taxid is: "

<< agp_taxid << endl;

274 if

(m_TaxidMap.size() == 1)

return true

;

276 if

(!use_xml) cerr <<

"Components with incorrect taxids:\n"

;

280 if

(map_it->first == agp_taxid)

continue

;

282 TTaxId

taxid = map_it->first;

285 out

" <CompBadTaxid taxid=\""

<< taxid

286

"\" line_num=\""

<< list_it->line_num

287

<< (list_it->file_num ?

string

(

290

"\">"

NStr::XmlEncode

(list_it->component_id)<<

"</CompBadTaxid>\n"

;

294 if

(list_it->file_num) {

295

cerr <<

->GetFile(list_it->file_num) <<

":"

;

297

cerr<< list_it->line_num <<

": "

<< list_it->component_id

298

" - Taxid "

<< taxid <<

"\n"

;

302 if

(!use_xml) cerr <<

"\n"

;

308 const string

& orig_line,

const string

& comp_id,

316

m_LineQueue.push_back(ld);

317

m_Accessions.insert(comp_id);

325

m_LineQueue.push_back(ld);

331

m_InvalidAccessions.clear();

332

vector<CSeq_id_Handle> idHandles;

334 for

(

const auto

& accession : m_Accessions) {

336

seqId.

Set

(accession);

340

m_InvalidAccessions.insert(accession);

349 for

(

auto

it=begin(bioseqHandles);

350

it!=end(bioseqHandles);

355 const auto

& bioseqHandle = *it;

357 const auto

& accVer = idHandle.GetSeqId()->IsGenbank() ?

358

idHandle.GetSeqId()->GetGenbank() :

359

idHandle.GetSeqId()->GetOther();

362 auto

& compInfo = m_ComponentInfoMap[accVer.GetAccession()];

363

compInfo.currentVersion = accVer.GetVersion();

364

compInfo.len = bioseqHandle.GetInst_Length();

366

compInfo.inDatabase=

true

;

368

m_Accessions.clear();

372 const string

& accession,

376 bool

versionSpecifiedInFile,

380 if

(versionSpecifiedInFile) {

381

ostr << line <<

'\n'

;

385 if

(currentVersion==1) {

388 for

(

size_t i

=0;

<line.

size

(); ++

) {

389 if

(line[

] ==

'\t'

) {

397

ostr << line.

substr

(0, pos);

398 if

(pos !=

) ostr <<

".1"

<< line.

substr

(pos);

403 if

(currentVersion) {

404

ostr << line <<

"#current version "

<< accession <<

"."

<< currentVersion <<

'\n'

;

409

ostr << line <<

"#component_id not in GenBank"

'\n'

;

418 for

(

const auto

& lineInfo : m_LineQueue)

421 const string

& orig_line = lineInfo.orig_line;

424

*m_pOut << orig_line <<

'\n'

;

426 pAgpErr

->LineDone(orig_line, lineInfo.line_num);

432 const bool

versionSpecified =

438

acc.

substr

(0, pos_ver) :

442 auto

itAccData = m_ComponentInfoMap.find(acc_nover);

444 if

(itAccData!=m_ComponentInfoMap.end()) {

445

acc_data = itAccData->second;

448 " - cannot get version for "

+acc);

453 " - cannot retrieve taxonomic id for "

+acc);

456

m_GenBankCompLineCount++;

462 if

(m_check_len_taxid) {

469 if

(m_SpeciesLevelTaxonCheck) {

470

taxid = x_GetSpecies(taxid);

472

x_AddToTaxidMap(taxid, acc, lineInfo.line_num);

477 auto

it = m_InvalidAccessions.find(acc);

478 if

(it != end(m_InvalidAccessions)) {

494 pAgpErr

->LineDone(orig_line, lineInfo.line_num);

502 if

(pos1==

)

return

long_acc;

504 SIZE_TYPE

pos2=long_acc.find(

'|'

, pos1+1);

505 if

(pos2==

)

return

long_acc.substr(pos1+1);

507 if

( long_acc.substr(0,pos1) ==

"gi"

) {

519 string

s = ids.front()->GetSeqIdString(

true

);

521

pos1 = s.rfind(

':'

);

522 if

( pos1 !=