RetroSearch Browse

<2; ++

)

168 if

(alignments.size() > 0)

169

display->

SetStartingColumn

(alignments.front()->GetFirstAlignedBlockPosition() - 5);

190

keepAlignments.push_back((*a)->Clone());

207 if

((*a)->NRows() != 2) {

208 ERRORMSG

(

"CreateSeqAlignFromUpdate() - can only save pairwise updates"

);

212

(*a)->GetUngappedAlignedBlocks(&

blocks

);

214 if

(!newSeqAlign)

continue

;

217 CUpdate_align

*updateAlign = (*a)->updateOrigin.GetPointer();

224 ERRORMSG

(

"UpdateViewer::SaveAlignments() - confused by Update-align format"

);

234

updateAlign->

SetType

((*a)->GetSequenceOfRow(1)->molecule ?

239

comment->

SetComment

(

"Created by demotion or import in Cn3D"

);

245

seqAnnotRef->

SetData

().SetAlign();

250 if

(!updateAlign || !(seqAlignList = &(updateAlign->

SetSeqannot

().SetData().SetAlign()))) {

251 ERRORMSG

(

"UpdateViewer::SaveAlignments() - can't find Update-align and/or Seq-align list"

);

257 if

(usedUpdateAligns.

find

(updateAlign) == usedUpdateAligns.

end

()) {

258

seqAlignList->clear();

259

updates.resize(updates.size() + 1);

260

updates.back().Reset(updateAlign);

261

usedUpdateAligns[updateAlign] =

true

;

265

seqAlignList->resize(seqAlignList->size() + 1);

266

seqAlignList->back().Reset(newSeqAlign);

288

vector < const Sequence * > chains;

290 if

(chains.size() == 1) {

293

wxString *titles =

new

wxString[chains.size()];

295 for

(choice=0; choice<(

)chains.size(); ++choice)

296

titles[choice] = chains[choice]->identifier->ToString().c_str();

297

choice = wxGetSingleChoiceIndex(

"Align to which protein chain?"

298 "Select Chain"

, chains.size(), titles);

300

master = chains[choice];

307 ERRORMSG

(

"UpdateViewer::GetMasterSequence() - no master sequence defined"

);

314

wxGetTextFromUser(

"Enter a list of protein GIs or Accessions:"

316 if

(ids.size() == 0)

return

;

318

wxStringTokenizer tkz(ids,

" ,;\t\r\n"

, wxTOKEN_STRTOK);

319 while

(tkz.HasMoreTokens()) {

320

wxString

id

= tkz.GetNextToken();

322 if

(bioseq.

Empty

())

327

newSequences->push_back(sequence);

329 ERRORMSG

(

"The sequence must be a protein"

);

336

newSequences->clear();

338

wxString fastaFile = wxFileSelector(

"Choose a FASTA file from which to import"

339 ""

""

"*.*"

, wxFD_OPEN | wxFD_FILE_MUST_EXIST, *

);

340 if

(fastaFile.size() > 0) {

345 ERRORMSG

(

"Unable to open file "

<< fastaFile.c_str());

358

newSequences->push_back(sequence);

362 if

(!(*b)->IsSeq()) {

363 WARNINGMSG

(

"CFastaReader returned nested Bioseq-set"

);

367

newSequences->push_back(sequence);

371 WARNINGMSG

(

"CFastaReader returned unknown format or no Bioseqs"

);

380

(*seqs)[1] = dependent;

384 ERRORMSG

(

"MakeEmptyAlignment() - error finalizing alignment"

);

394

newAlignments->clear();

395

SequenceList::const_iterator s, se = newSequences.end();

396 for

(s=newSequences.begin(); s!=se; ++s) {

398 if

(newAlignment) newAlignments->push_back(newAlignment);

405 static const

wxString choiceStrings[] = {

"Network via GI/Accession"

"From a FASTA File"

};

406 enum

choiceValues { FROM_GI=0, FROM_FASTA, N_CHOICES };

407 int

importFrom = wxGetSingleChoiceIndex(

"From what source would you like to import sequences?"

408 "Select Import Source"

, N_CHOICES, choiceStrings, *

);

409 if

(importFrom < 0)

return

;

412 if

(importFrom == FROM_GI)

416 else if

(importFrom == FROM_FASTA)

424 if

(!master)

return

;

429 if

(newSequences.size() == 0) {

430 WARNINGMSG

(

"UpdateViewer::ImportSequence() - no sequences were imported"

);

439 if

(newAlignments.size() > 0)

442 ERRORMSG

(

"UpdateViewer::ImportSequence() - no new alignments were created"

);

448 unsigned int

masterFrom,

unsigned int

masterTo)

const 451 WARNINGMSG

(

"UpdateViewer::GetVASTAlignments() - " 456

SequenceList::const_iterator s, se = newSequences.end();

457 for

(s=newSequences.begin(); s!=se; ++s) {

458 if

((*s)->identifier->pdbID.size() == 0) {

459 WARNINGMSG

(

"UpdateViewer::GetVASTAlignments() - " 460 "can't be called with non-MMDB dependent "

<< (*s)->identifier->ToString());

462 if

(newAlignment) newAlignments->push_back(newAlignment);

468

host =

"www.ncbi.nlm.nih.gov"

469

path =

"/Structure/VA/vastalign.cgi"

, err;

472

"&neighbor="

<< (*s)->identifier->ToString();

473 if

(masterFrom <= masterTo && masterFrom < master->Length() && masterTo < master->Length())

474

argstr <<

"&from="

<< (masterFrom+1) <<

"&to="

<< (masterTo+1);

479 INFOMSG

(

"trying to load VAST alignment data from "

<< host << path <<

'?'

<< args);

482 ERRORMSG

(

"Error calling vastalign.cgi: "

<< err);

484 if

(newAlignment) newAlignments->push_back(newAlignment);

487 INFOMSG

(

"successfully loaded data from vastalign.cgi"

);

489 WriteASNToFile

(

"vastalign.dat.txt"

, structureAlignment,

false

, &err);

500 if

(structureAlignment.

IsSetId

() && structureAlignment.

().front()->IsMmdb_id() &&

501

structureAlignment.

().front()->GetMmdb_id().Get() == 0)

503 WARNINGMSG

(

"VAST found no alignment for these chains"

);

508 if

(!structureAlignment.

IsSetId

() || !structureAlignment.

().front()->IsMmdb_id() ||

510

structureAlignment.

().front()->GetFeatures().size() != 1 ||

511

!structureAlignment.

().front()->GetFeatures().front()->IsSetLocation() ||

512

!structureAlignment.

().front()->GetFeatures().front()->GetLocation().IsAlignment())

514 throw "VAST data does not contain exactly one alignment of recognized format - " 515 "possibly a problem with vastalign.cgi"

;

519 throw "Master structure MMDB ID mismatch - check to see if this structure been updated"

;

523

structureAlignment.

().front()->GetFeatures().front()->GetLocation().GetAlignment();

529

alignment.

GetBiostruc_ids

().back()->GetMmdb_id().Get() != (*s)->identifier->mmdbID ||

531

!alignment.

().front()->GetResidues().IsInterval() ||

533

!alignment.

().back()->GetResidues().IsInterval() ||

534

alignment.

().front()->GetResidues().GetInterval().size() !=

535

alignment.

().back()->GetResidues().GetInterval().size())

537 throw "Unrecognized VAST data format"

;

541

CResidue_pntrs::TInterval::const_iterator

, j,

542

ie = alignment.

().front()->GetResidues().GetInterval().end();

543 for

(

=alignment.

().front()->GetResidues().GetInterval().begin(),

544

j=alignment.

().back()->GetResidues().GetInterval().begin();

!=ie; ++

, ++j)

547

(*j)->GetMolecule_id().Get() != (*s)->identifier->moleculeID)

549 throw "Mismatch in molecule ids in alignment interval block"

;

552

newBlock->

SetRangeOfRow

(0, (*i)->GetFrom().Get() - 1, (*i)->GetTo().Get() - 1);

553

newBlock->

SetRangeOfRow

(1, (*j)->GetFrom().Get() - 1, (*j)->GetTo().Get() - 1);

554

newBlock->

width

= (*i)->GetTo().Get() - (*i)->GetFrom().Get() + 1;

561

structureAlignments->resize(structureAlignments->size() + 1);

562

structureAlignments->back().structureAlignment =

563

structureAlignment.

SetFeatures

().front()->SetFeatures().front();

564

structureAlignments->back().masterDomainID =

565

structureAlignment.

().front()->GetId().Get();

566

structureAlignments->back().dependentDomainID =

567

structureAlignment.

().front()->GetFeatures().front()->GetId().Get();

569 throw "No structure alignment in VAST data blob"

;

571

}

catch

(

const char

*err) {

572 ERRORMSG

(

"Failed to import VAST alignment: "

<< err);

577 ERRORMSG

(

"MakeEmptyAlignment() - error finalizing alignment"

);

581

newAlignments->push_back(newAlignment);

589 if

(!master)

return

;

591 ERRORMSG

(

"Can't import another structure when current structure has multiple models"

);

596 static const

wxString choiceStrings[] = {

"Via Network"

"From a File"

};

597 enum

choiceValues { FROM_NETWORK=0, FROM_FILE, N_CHOICES };

598 int

importFrom = wxGetSingleChoiceIndex(

599 "From what source would you like to import the structure?"

"Select Import Source"

601 if

(importFrom < 0)

return

;

606 if

(importFrom == FROM_NETWORK) {

607

wxString

id

= wxGetTextFromUser(

"Enter a PDB or MMDB ID:"

"Input Identifier"

""

, *

);

612

biostruc, &bioseqs)) {

613 ERRORMSG

(

"Failed to load structure "

id

.c_str());

618 else if

(importFrom == FROM_FILE) {

619 string

filename =

WX_TO_STD

(wxFileSelector(

"Choose a single-structure file:"

621 if

(filename.size() == 0)

return

;

622 bool

readOK =

false

;

624 TRACEMSG

(

"trying to read file '"

<< filename <<

"' as binary mime"

);

631 TRACEMSG

(

"trying to read file '"

<< filename <<

"' as ascii mime"

);

639 ERRORMSG

(

"Couldn't read structure from "

<< filename);

647 if

(biostruc->

().size() == 0 || !biostruc->

().front()->IsMmdb_id()) {

648 ERRORMSG

(

"Can't get MMDB ID from loaded structure"

);

651

mmdbID = biostruc->

().front()->GetMmdb_id().Get();

655 if

(biostruc->

GetModel

().size() != 1 ||

660 ERRORMSG

(

"Biostruc does not match current data - should be " 661

<< (master->

parentSet

isAlphaOnly

"alpha-only"

"one-coordinate-per-atom"

) <<

" model"

);

667

vector < pair < const CSeq_id * , string > > chains;

669

CBiostruc_graph::TDescr::const_iterator d, de;

670

CBiostruc_graph::TMolecule_graphs::const_iterator

673 bool

isProtein =

false

;

676 string

full_name =

""

;

679

de = (*m)->GetDescr().end();

680 for

(d=(*m)->GetDescr().begin(); d!=de; ++d) {

681 if

((*d)->IsName()) {

682

full_name = (*d)->GetName();

683

name = (*d)->GetName()[0];

685 else if

((*d)->IsMolecule_type() &&

688 if

(isProtein && (full_name !=

""

))

break

;

692 if

((*m)->IsSetSeq_id())

693

sid = &((*m)->GetSeq_id());

696 if

(isProtein && (full_name !=

""

) && sid !=

NULL

) {

697

moleculeIDs[sid] = (*m)->GetId().Get();

698

chains.push_back(make_pair(sid, full_name));

701 if

(chains.size() == 0) {

702 ERRORMSG

(

"No protein chains found in this structure!"

);

708

CBiostruc::TDescr::const_iterator

= biostruc->

GetDescr

().end();

710 if

((*n)->IsName()) {

711

pdbID = (*n)->GetName();

717

vector < const CSeq_id * > sids;

718

vector < string > sids_chain_ids;

719 if

(chains.size() == 1) {

720

sids.push_back(chains[0].

first

);

721

sids_chain_ids.push_back(chains[0].second);

723

wxString *choices =

new

wxString[chains.size()];

725 for

(choice=0; choice<(

)chains.size(); ++choice)

726

choices[choice].Printf(

"%s_%s %s"

727

pdbID.c_str(), chains[choice].second, chains[choice].first->GetSeqIdString().c_str());

728

wxArrayInt selections;

730 int

nsel = wxGetSelectedChoices(selections,

"Which chain(s) do you want to align?"

731 "Select Chain"

, chains.size(), choices, *

);

732 if

(nsel == 0)

return

;

733 for

(choice = 0; choice < nsel; ++choice) {

734

sids.push_back(chains[selections[choice]].

first

);

735

sids_chain_ids.push_back(chains[selections[choice]].second);

742 for

(

unsigned int

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

746 if

((*s)->identifier->MatchesSeqId(*(sids[j]))) {

748

newSequences.push_back(*s);

749

seq2id[*s] = sids[j];

756

BioseqRefList::iterator

, be = bioseqs.end();

757 for

(

=bioseqs.begin();

!=be; ++

) {

758

CBioseq::TId::const_iterator

, ie = (*b)->GetId().end();

759 for

(

=(*b)->GetId().begin();

!=ie; ++

) {

760 if

((*i)->Match(*(sids[j])))

811

newSequences.push_back(sequence);

812

seq2id[sequence] = sids[j];

824

SequenceList::const_iterator w, we = newSequences.end();

825 for

(w=newSequences.begin(); w!=we; ++w) {

828

(*w)->AddMMDBAnnotTag(mmdbID);

834 if

((*w)->identifier->mmdbID != mmdbID)

835 ERRORMSG

(

"MMDB ID mismatch in sequence "

<< (*w)->identifier->ToString()

836

"; "

<< (*w)->identifier->mmdbID <<

" vs "

<< mmdbID);

838 if

(moleculeIDs.

find

(seq2id[*w]) != moleculeIDs.

end

()) {

841

moleculeIDs[seq2id[*w]];

843 if

((*w)->identifier->moleculeID != moleculeIDs[seq2id[*w]])

844 ERRORMSG

(

"Molecule ID mismatch in sequence "

<< (*w)->identifier->ToString());

847 ERRORMSG

(

"No matching id for MMDB sequence "

<< (*w)->identifier->ToString());

853 int

masterFrom = -1, masterTo = -1;

858 if

(aBlocks.size() > 0) {

859

masterFrom = aBlocks.front()->GetRangeOfRow(0)->from;

860

masterTo = aBlocks.back()->GetRangeOfRow(0)->to;

866

SequenceList::const_iterator s, se = newSequences.end();

867 for

(s=newSequences.begin(); s!=se; ++s) {

876 ERRORMSG

(

"ImportStructure() - error finalizing alignment"

);

880

newAlignments.push_back(newAlignment);

885 if

(newAlignments.size() == newSequences.size())

888 ERRORMSG

(

"UpdateViewer::ImportStructure() - no new alignments were created"

);

896

wxMessageBox(

"The structure has been successfully imported! However, it will not appear until you\n" 897 "save this data to a file and then re-load it in a new session. And depending on the type\n" 898 "of data, it still might not appear unless the corresponding new pairwise alignment has\n" 899 "been merged into the multiple alignment."

900 "Structure Added"

, wxOK | wxICON_INFORMATION, *

);

903

wxMessageBox(

"The selected sequence from this structure has been imported! However, the structure\n" 904 "itself can not appear in the structure window until the sequence has been merged into the\n" 905 "multiple alignment and the alignment has been remastered to a structured sequence."

906 "Sequence Added"

, wxOK | wxICON_INFORMATION, *

);

912 TRACEMSG

(

"saving pending imported structures and structure alignments"

);

920 ERRORMSG

(

"UpdateViewer::SavePendingStructures() - error saving Biostruc"

);

936 if

(usePSSMFromMultiple && !multipleForPSSM) {

937 ERRORMSG

(

"Can't do BLAST/PSSM when no multiple alignment is present"

);

944 if

!= alignment)

continue

;

948

toRealign.push_back(alignment);

951

multipleForPSSM, toRealign, &newAlignments, usePSSMFromMultiple);

952 if

(newAlignments.size() != 1) {

953 ERRORMSG

(

"UpdateViewer::BlastUpdate() - CreateNewPairwiseAlignmentsByBlast() failed"

);

956 if

(newAlignments.front()->NAlignedBlocks() == 0) {

958 delete

newAlignments.front();

963 TRACEMSG

(

"BLAST succeeded - replacing alignment"

);

965

= newAlignments.front();

978 int

dependentRow, vector < int > *dependent2master)

980

dependent2master->clear();

984

BlockMultipleAlignment::UngappedAlignedBlockList::const_iterator

, be = uaBlocks.end();

985 for

(

=uaBlocks.begin();

!=be; ++

) {

987

*masterRange = (*b)->GetRangeOfRow(0),

988

*dependentRange = (*b)->GetRangeOfRow(dependentRow);

989 for

(

unsigned int i

=0;

<(*b)->width; ++

)

990

(*dependent2master)[dependentRange->from +

] = masterRange->

from

;

998 if

(rowAlignments.size() != multiple->

NRows

()) {

999 ERRORMSG

(

"GetAlignmentByBestNeighbor: wrong # alignments"

);

1005 unsigned int b

, bestRow = 0;

1006

BLASTer::AlignmentList::const_iterator p, pe = rowAlignments.end();

1007 for

(

=0, p=rowAlignments.begin(); p!=pe; ++

, ++p) {

1009

bestMatchFromMultiple = *p;

1013 if

(!bestMatchFromMultiple || bestMatchFromMultiple->

GetRowDouble

(0) > 0.000001) {

1014 WARNINGMSG

(

"GetAlignmentByBestNeighbor: no significant hit found"

);

1017 INFOMSG

(

"Closest neighbor from multiple: sequence " 1019

", E-value: "

<< bestMatchFromMultiple->

GetRowDouble

(0));

1025 if

(bestRow == 0)

return

bestMatchFromMultiple->

Clone

();

1028

vector < int > import2dependent, dependent2master;

1035

(*seqs)[1] = importSeq;

1040 int

masterStart=-1, importStart, importLoc, dependentLoc, masterLoc,

len

=0;

1041 for

(importStart=-1, importLoc=0; importLoc<=(

)importSeq->

Length

(); ++importLoc) {

1044

dependentLoc = (importLoc < (

)importSeq->

Length

()) ? import2dependent[importLoc] : -1;

1045

masterLoc = (dependentLoc >= 0) ? dependent2master[dependentLoc] : -1;

1048 if

(importStart >= 0) {

1051 if

(masterLoc >= 0 && masterLoc-masterStart == importLoc-importStart) {

1067 if

(importStart < 0) {

1068 if

(masterLoc >= 0) {

1069

masterStart = masterLoc;

1070

importStart = importLoc;

1078 ERRORMSG

(

"error finalizing alignment"

);

1079 delete

newAlignment;

1083 return

newAlignment;

1090 ERRORMSG

(

"Can't do BLAST Neighbor when no multiple alignment is present"

);

1095 if

(uaBlocks.size() == 0) {

1096 ERRORMSG

(

"Can't do BLAST Neighbor with null multiple alignment"

);

1104 if

== update)

break

;

1113

(*seqs)[1] = updateSeq;

1120 WARNINGMSG

(

"Can't get footprint excess residues from registry"

);

1121

newAlignment->

alignMasterFrom

= uaBlocks.front()->GetRangeOfRow(

)->from - excess;

1124

newAlignment->

alignMasterTo

= uaBlocks.back()->GetRangeOfRow(