RetroSearch Browse

) {

146

reverse(mismatches.begin(),mismatches.end());

147

reverse(mismstatus.begin(),mismstatus.end());

152 bool

ggap_model=

false

;

159 if

(is_product_reversed) {

160 int tmp

= prod_cur_start;

161

prod_cur_start = product_len - prod_cur_end -1;

162

prod_cur_end = product_len -

tmp

-1;

170 if

(prod_prev+1 != prod_cur_start || !prev_3_prime_splice || !cur_5_prime_splice) {

172 if

(!mismatches.empty())

173

mismatches = mismatches.substr(prod_cur_start - prod_prev -1);

186 if

(Strand() ==

) {

194 if

((*it)->CanGetId() && (*it)->GetId().IsStr()) {

195 if

((*it)->GetId().GetStr() ==

"idty"

) {

196

eident = (*it)->GetValue().GetReal();

213 string

fill_seq = transcript.substr(nuc_cur_start, nuc_cur_end-nuc_cur_start+1);

225

AddGgapExon(eident, fill_seq, fill_src, Strand() ==

);

232 _ASSERT

(transcript_exons.back().NotEmpty());

240 if

(mismstatus[0] ==

'n'

)

242 else if

(mismstatus[0] ==

'c'

)

244

mismstatus = mismstatus.substr(1);

249 if

(!mismatches.empty()) {

250

v = mismatches.substr(0,product_ins);

251

mismatches = mismatches.substr(product_ins);

254

reverse(v.begin(),v.end());

256 if

(Strand() ==

)

257

indels.push_back(fs);

259

indels.insert(indels.begin(), fs);

262 if

(Strand()==

)

267

}

else if

(chunk.

IsMatch

()) {

271 string

v(mismatch_len,

'N'

);

272 if

(!mismatches.empty()) {

273 _ASSERT

(mismatch_len <= (

int

)mismatches.length());

274

v = mismatches.substr(0,mismatch_len);

275

mismatches = mismatches.substr(mismatch_len);

277 if

(Strand() ==

) {

280

reverse(v.begin(),v.end());

288 if

(Strand() ==

)

289

indels.back().SetStatus(indelstatus);

291

indels.front().SetStatus(indelstatus);

295

prod_prev = prod_cur_end;

298 _ASSERT

(mismatches.empty() || (product_len - prod_prev - 1 == (

int

)mismatches.length()));

300 sort

(transcript_exons.begin(),transcript_exons.end());

301 bool

minusstrand = Strand() ==

;

303 if

(orientation ==

)

304

reverse(transcript_exons.begin(),transcript_exons.end());

307 _ASSERT

(orientation == Strand());

310 if

((*m)->IsStop_codon_found()) {

312 if

(Strand() ==

) {

314 _ASSERT

((transcript_exons.back().GetTo()+1)%3 == 0);

315

transcript_exons.back().SetTo(transcript_exons.back().GetTo()+3);

318 _ASSERT

((transcript_exons.front().GetTo()+1)%3 == 0);

319

transcript_exons.front().SetTo(transcript_exons.front().GetTo()+3);

327 bool

keepdoing =

true

;

331

TInDels::iterator indl_next = indl;

332 if

(++indl_next == indels.end())

335 if

(indl->InDelEnd() == indl_next->Loc()) {

336 string

new_seq = indl->GetInDelV()+indl_next->GetInDelV();

338 if

(indl->GetType() == indl_next->GetType()) {

339

*indl =

CInDelInfo

(indl->Loc(), indl->Len()+indl_next->Len(), indl->GetType(), new_seq);

340

indels.erase(indl_next);

342

}

else if

(!indl->IsMismatch() && !indl_next->IsMismatch()) {

343 if

(indl->Len() == indl_next->Len()) {

345

indels.erase(indl_next);

346

}

else if

(indl->Len() < indl_next->Len()) {

348

*indl_next =

CInDelInfo

(indl->InDelEnd(), indl_next->Len()-indl->Len(), indl_next->GetType(), new_seq.substr(indl->Len()));

350

*indl =

CInDelInfo

(indl->Loc(), indl->Len()-indl_next->Len(), indl->GetType(), new_seq.substr(0,new_seq_len-indl_next->Len()));

360 if

(!indels.empty()) {

361 auto

& indl = indels.front();

362 auto

& e = Exons().front();

363 if

(!indl.IsMismatch() && indl.Loc() == e.GetFrom()) {

364 if

(indl.IsInsertion()) {

365

e.AddFrom(indl.Len());

368 auto

= transcript_exons.front();

369 if

(orientation ==

)

370 t

.SetFrom(

.GetFrom()+indl.Len());

372 t

.SetTo(

.GetTo()-indl.Len());

374

indels.erase(indels.begin());

377 if

(!indels.empty()) {

378 auto

& indl = indels.back();

379 auto

& e = Exons().back();

380 if

(!indl.IsMismatch() && indl.InDelEnd() == e.GetTo()+1) {

381 if

(indl.IsInsertion()) {

382

e.AddTo(-indl.Len());

385 auto

= transcript_exons.back();

386 if

(orientation ==

)

387 t

.SetTo(

.GetTo()-indl.Len());

389 t

.SetFrom(

.GetFrom()+indl.Len());

394

Exons().front().m_fsplice_sig.clear();

395

Exons().back().m_ssplice_sig.clear();

397

m_alignmap =

CAlignMap

(Exons(), transcript_exons, indels, orientation, target_len );

398

FrameShifts() = indels;

407 for

(CGeneModel::TExons::const_iterator piece_begin = Exons().begin(); piece_begin != Exons().end(); ++piece_begin) {

408 _ASSERT

( !piece_begin->m_fsplice );

410 if

(piece_begin->Limits().Empty()) {

411 _ASSERT

(piece_begin->m_ssplice);

413 _ASSERT

(piece_begin->Limits().NotEmpty());

416

CGeneModel::TExons::const_iterator piece_end;

417 for

(piece_end = piece_begin; piece_end != Exons().end() && piece_end->m_ssplice; ++piece_end) ;

418 _ASSERT

( piece_end != Exons().end() );

420

CGeneModel::TExons::const_iterator piece_end_g = piece_end;

421 if

(piece_end_g->Limits().Empty()) {

422 _ASSERT

(piece_end_g->m_fsplice);

424 _ASSERT

(piece_end_g->Limits().NotEmpty());

427 TSignedSeqRange

piece_range(piece_begin->GetFrom(),piece_end_g->GetTo());

429

piece_range = m_alignmap.ShrinkToRealPoints(piece_range, is_protein);

446 _ASSERT

(piece_range.NotEmpty());

447 _ASSERT

(piece_range.IntersectingWith(piece_begin->Limits()) && piece_range.IntersectingWith(piece_end_g->Limits()));

449 if

(piece_range.GetFrom() != piece_begin->GetFrom() || piece_range.GetTo() != piece_end_g->GetTo()) {

454

piece_begin = piece_end;

458 TSignedSeqRange

reading_frame = m_alignmap.MapRangeOrigToEdited(Limits(),

true

);

462 if

((*m)->IsStart_codon_found()) {

465

}

else if

((*m)->IsStop_codon_found()) {

473

cds_info_t.SetReadingFrame(reading_frame,

true

);

475

cds_info_t.SetStart(start,

false

);

478

cds_info_t.SetStop(stop,

false

);

482

SetCdsInfo(cds_info_g);

484

SetCdsInfo(cds_info_t);

494 if

((*it)->CanGetId() && (*it)->GetId().IsStr()) {

495 string

scr = (*it)->GetId().GetStr();

496 if

((scr ==

"N of matches"

) || (scr ==

"num_ident"

) || (scr ==

"matches"

)) {

497 double

ident = (*it)->GetValue().GetInt();

500

}

else if

(scr ==

"rank"

&& (*it)->GetValue().GetInt() == 1) {

502

}

else if

(scr ==

"ambiguous_orientation"

) {

504

}

else if

(scr ==

"count"

) {

505 _ASSERT

(Weight() == 1 || Weight() == (*it)->GetValue().GetInt());

506

SetWeight((*it)->GetValue().GetInt());

530 string

cds_seq(cds_len,

'A'

);

531 copy

(mrna.begin()+cds_start, mrna.begin()+cds_start+cds_len, cds_seq.begin());

554

prot_seq[(stp->GetFrom()- cds_info.

Cds

().

GetFrom

())/3] =

'U'

;

567 if

(prot_seq[0] ==

'-'

) {

568 string

first_triplet = cds_seq.substr(0, 3);

571

prot_seq = first_aa+prot_seq.substr(1);

582

prot_seq[(stp->GetFrom()- cds_info.

Cds

().

GetFrom

())/3] =

'U'

;

594 CRef<CSeq_loc>

s_ExonDataToLoc(

const

vector<TSignedSeqRange>& vec,

600 ITERATE

(vector<TSignedSeqRange>, iter, vec) {

602

ival->SetFrom(iter->GetFrom());

603

ival->SetTo(iter->GetTo());

604

ival->SetStrand(strand);

605

ival->SetId().Assign(

id

);

607 data

.push_back(ival);

610 if

(

data

.size() == 1) {

611

loc->SetInt(*

data

.front());

613

loc->SetPacked_int().Set().swap(

data

);

627

annot->SetNameDesc(

"Gnomon gene scan output"

);

631 unsigned int

counter = 0;

632 string

locus_tag_base(

"GNOMON_"

);

635 int

strand = igene.

Strand

();

638

vector<TSignedSeqRange> mrna_vec;

639 copy

(igene.

Exons

().begin(), igene.

Exons

().end(), back_inserter(mrna_vec));

640

vector<TSignedSeqRange> cds_vec;

642 for

(

size_t

j = 0; j < mrna_vec.size(); ++j) {

644 if

(!intersect.

Empty

()) {

645

cds_vec.push_back(intersect);

651 if

(strand ==

) {

652 _ASSERT

(cds_vec.back().GetLength()>=3);

653

cds_vec.back().SetTo(cds_vec.back().GetTo() - 3);

655 _ASSERT

(cds_vec.front().GetLength()>=3);

656

cds_vec.front().SetFrom(cds_vec.front().GetFrom() + 3);

663 if

(mrna_vec.size()) {

667

(*s_ExonDataToLoc(mrna_vec,

676 if

(!cds_vec.empty()) {

681

(*s_ExonDataToLoc(cds_vec,

690

snprintf(

buf

, 32,

"%04u"

, ++counter);

691 string

name(locus_tag_base);

693

feat_gene->

SetData

().SetGene().SetLocus_tag(name);

704 ftable

.push_back(feat_gene);

705 ftable

.push_back(feat_mrna);

707 ftable

.push_back(feat_cds);

731 unsigned int

gc_count = 0;

733 for

( ; xcript_iter; ++xcript_iter) {

734 if

(*xcript_iter ==

'G'

|| *xcript_iter ==

'C'

) {

738

*gccontent =

static_cast<unsigned int>

(100.0 * gc_count / xcript_vec.

size

() + 0.5);

745

vec.SetIupacCoding();

747

seq.reserve(vec.size());

749 for

( ; iter; ++iter) {

756 for

(

unsigned int i

= 5;

< seq.size() - 3; ++

)

757

score += cdr.

(seq,

% 3) - ncdr.

(seq,

);

767 int

totallen = xcript_vec.

size

();

770 int

extrabases = start.

Left

()+2;

773 if

(startposition < extrabases) {

775

extraNs5p = extrabases-startposition;

777

left = startposition-extrabases;

779

right =

min

(startposition+2+start.

Right

(),totallen-1);

782 if

(startposition+extrabases >= totallen) {

784

extraNs5p = startposition+extrabases-totallen+1;

786

right = startposition+extrabases;

788

left =

max

(0,startposition-2-start.

Right

());

796 id

->Assign(*seq_id);

804

sttseq.resize(extraNs5p,

enN

);

805 for

(

unsigned int i

= 0;

< sttvec.

size

(); ++

) {

806

sttseq.push_back(

fromACGT

(sttvec[

]));

810

*startscore = start.

(sttseq, extrabases+2);

812 for

(

unsigned int i

= 5;

< sttseq.size(); ++

) {

813

*startscore -= ncdr.