RetroSearch Browse

();

231

}

else if

(denseg.

().size() < dim) {

232

dim = (unsigned)denseg.

().size();

244 if

((*it)->IsSetId() && (*it)->GetId().IsStr() &&

246

(*it)->IsSetValue() && (*it)->GetValue().IsReal()) {

247 if

((*it)->GetValue().GetReal() > 50.0) {

260 int

dim = denseg.

();

265 bool

is_tpa =

false

;

271 while

(desc_ci && !is_tpa) {

288 bool

is_tpa =

false

;

294 while

(desc_ci && !is_tpa) {

309 int

dim = denseg.

();

325 while

(col < aln_len && !ids_missing) {

337

string::const_iterator it1 =

.begin();

338

string::const_iterator it2 = it1;

345 if

(it2 ==

.end()) {

368

}

catch

(std::exception &) {

392 if

((*id)->IsGenbank() && (*id)->GetGenbank().IsSetVersion() && (*id)->GetGenbank().GetVersion() == 0) {

406 if

((*id)->IsGenbank() && (*id)->GetGenbank().IsSetVersion() && (*id)->GetGenbank().GetVersion() == 0) {

407

(*id)->SetGenbank().ResetVersion();

416 size_t

num_match = 0;

417 bool

ids_missing =

false

;

426 int

dim = denseg.

();

434 tmp

->Assign(denseg);

486

vector <string> aln_rows;

487

vector <TSeqPos> row_starts;

488

vector <TSeqPos> row_stops;

497

aln_rows.push_back (sequence);

500

row_starts.push_back (start);

504

}

catch

(std::exception &) {

511 bool

any_data =

false

;

514 while

(col < aln_len) {

518 if

(row_starts[

] >= col && row_stops[

] <= col

519

&& aln_rows[

].length() > col) {

520 string

nt = aln_rows[

].substr(col - row_starts[

], 1);

536

string::const_iterator it1 =

.begin();

537

string::const_iterator it2 = it1;

544 if

(it2 ==

.end()) {

560

}

catch

(std::exception &) {

572 size_t

pct_id = (num_match * 100) / col;

575 "PercentIdentity: This alignment has a percent identity of "

NStr::NumericToString

(pct_id) +

"%"

589 size_t

dim = denseg.GetDim();

590 size_t

numseg = denseg.GetNumseg();

592

denseg.GetIds()[0]->GetLabel (&

);

596 if

( bar_pos != string::npos ) {

604 if

( dim != denseg.GetIds().size() ) {

606 "SeqId: The Seqalign has more or fewer ids than the number of rows in the alignment (context " 607

"). Look for possible formatting errors in the ids."

, align);

611 if

( numseg != denseg.GetLens().size() ) {

613 "Mismatch between specified numseg ("

615 ") and number of Lens ("

621 if

( dim * numseg != denseg.GetStarts().size() ) {

623 "The number of Starts ("

625 ") does not match the expected size of dim * numseg ("

648 size_t

dim = packed.GetDim();

649 size_t

numseg = packed.GetNumseg();

652 if

( dim != packed.GetIds().size() ) {

654 "SeqId: The Seqalign has more or fewer ids than the number of rows in the alignment. Look for possible formatting errors in the ids."

, align);

658 if

( numseg != packed.GetLens().size() ) {

660 "Mismatch between specified numseg ("

662 ") and number of Lens ("

680 size_t

num_dendiag = 0;

685 size_t

dim = dendiag.

();

694 if

( bar_pos != string::npos ) {

701 if

( dim != dendiag.

().size() ) {

704

", there are more or fewer rows than there are seqids (context " 705

"). Look for possible formatting errors in the ids."

, align);

709 if

( dim != dendiag.

GetStarts

().size() ) {

711 "Mismatch between specified dimension ("

713 ") and number ofStarts ("

722 "Mismatch between specified dimension ("

724 ") and number of Strands ("

742 const TStd

& std_segs,

745 size_t

num_stdseg = 0;

749 const CStd_seg

& stdseg = **stdseg_iter;

750 size_t

dim = stdseg.

();

756 if

( dim != stdseg.

GetLoc

().size() ) {

761 if

( bar_pos != string::npos ) {

768

", there are more or fewer rows than there are seqids (context " 769

"). Look for possible formatting errors in the ids."

, align);

774 if

( dim != stdseg.

().size() ) {

776 "Mismatch between specified dimension ("

778 ") and number of Seq-ids ("

795 template

typename

803 if

( !obj.IsSetDim() || obj.GetDim() == 0) {

809 "Dim: This alignment has dimension zero"

, align);

811

}

else if

(obj.GetDim() == 1) {

816 msg

"Segs: Segment "

NStr::SizetToString

(part) +

" apparently has only one sequence. Each portion of the alignment must have at least two sequences."

;

819 msg

"Dim: This seqalign apparently has only one sequence. Each alignment must have at least two sequences."

;

850 if

( !denseg.IsSetStrands() ) {

854 size_t

dim = denseg.GetDim();

855 size_t

numseg = denseg.GetNumseg();

859 for

(

size_t id

= 0;

id

< dim; ++id ) {

862 for

(

size_t

seg = 0; seg < numseg; ++seg ) {

863 ENa_strand

strand2 = strands[

id

+ (seg * dim)];

878 if

( strand1 != strand2 ) {

881 if

(denseg.GetIds().size() >

id

&& denseg.GetIds()[id]) {

882

sid = denseg.GetIds()[id]->AsFastaString();

884 if

(denseg.GetStarts().size() >

id

+ (seg * dim)) {

885

pos = denseg.GetStarts()[

id

+ (seg * dim)];

888 "Strand: The strand labels for SeqId "

890 " are inconsistent across the alignment. " 891 "The first inconsistent region is the "

902 const TStd

& std_segs,

918 string

id_label =

id

->AsFastaString();

929

strands[id_label] = strand;

930

reported[id_label] =

false

;

931

}

else if

(!reported[id_label]

932

&& strands[id_label] != strand ) {

935 "Strand: The strand labels for SeqId "

+ id_label +

936 " are inconsistent across the alignment. The first inconsistent region is the " 939

reported[id_label] =

true

;

959

}

else if

(!b1.

() && b2.

()) {

966 for

(

CSeqVector_CI

sv1_iter(sv1), sv2_iter(sv2); (sv1_iter) && (sv2_iter); ++sv1_iter, ++sv2_iter ) {

967 if

(*sv1_iter == *sv2_iter || *sv1_iter ==

'N'

|| *sv2_iter ==

'N'

) {

995

denseg.GetDim() <= 2 ) {

999 size_t

dim = denseg.GetDim();

1000 size_t

numseg = denseg.GetNumseg();

1002

vector<string> fasta_like;

1008 for

(

size_t

seg = 0; seg < numseg; ++ seg ) {

1010 if

( starts[

id

+ (dim * seg)] < 0 ) {

1019 if

( seg == numseg - 1) {

1022

fasta_like.push_back(denseg.GetIds()[id]->AsFastaString());

1027 if

( !fasta_like.empty() ) {

1028

CDense_seg::TIds::const_iterator id_it = denseg.GetIds().begin();

1030 string context

= (*id_it)->GetSeqIdString();

1032 bool

is_fasta_like =

false

;

1035 while

(id_it != denseg.GetIds().end() && !is_fasta_like) {

1038

is_fasta_like =

true

;

1043

is_fasta_like =

true

;

1045 if

(is_fasta_like) {

1047 "Fasta: This may be a fasta-like alignment for SeqId: " 1048

+ fasta_like.front() +

" in the context of "

, align);

1058 size_t

align_pos = 0;

1060 int

numseg = denseg.GetNumseg();

1061 int

dim = denseg.GetDim();

1064 for

(

size_t

seg = 0; seg < numseg; ++seg) {

1065 bool

seggap =

true

;

1066 for

(

int id

= 0;

id

< dim && seg * dim +

id

< starts.size(); ++id) {

1067 if

(starts[seg * dim +

id

] != -1) {

1075 if

(denseg.IsSetIds() && denseg.GetIds().size() > 0) {

1083 if

(denseg.IsSetLens() && denseg.GetLens().size() > (

unsigned int

)seg) {

1084

align_pos += denseg.GetLens()[seg];

1097

+ itr->label +

" contains only gaps. Each segment must contain at least one actual sequence -- look for columns with all gaps and delete them."

1122 static Uchar

bits[] = { 0x80, 0x40, 0x20, 0x10, 0x08, 0x04, 0x02, 0x01 };

1124 size_t

numseg = packed.GetNumseg();

1125 size_t

dim = packed.GetDim();

1128 size_t

align_pos = 0;

1129 for

(

size_t

seg = 0; seg < numseg; ++seg) {

1131 for

(;

id

< dim; ++id) {

1132 size_t i

id

+ (dim * seg);

1133 if

((present[

/ 8] & bits[

% 8])) {

1140 if

(packed.IsSetIds() && packed.GetIds().size() > 0) {

1148 if

(packed.IsSetLens() && packed.GetLens().size() > seg) {

1149

align_pos += packed.GetLens()[seg];

1171 size_t

align_pos = 0;

1177 if

(!(*loc)->IsEmpty() && !(*loc)->IsNull()) {

1180

}

else if

(

len

== 0) {

1201

(

const TStd

& std_segs,

1216 if

(!(*diag_seg)->IsSetDim() || (*diag_seg)->GetDim() == 0) {

1218 if

((*diag_seg)->IsSetIds() && (*diag_seg)->GetIds().size() > 0) {

1219

(*diag_seg)->GetIds().front()->GetLabel(&

);

1226 if

((*diag_seg)->IsSetLen()) {

1227

align_pos += (*diag_seg)->GetLen();

1253

vector< CRef< CSeq_id > > ids;

1260 const CSeq_id

id

= **id_iter;

1261 if

(

id

.IsLocal() ) {

1264 if

( hdl.

IsNa

() ) {

1267 if

( hdl.

() ) {

1273 "SeqId: The sequence corresponding to SeqId "

1274 id

.AsFastaString() +

" could not be found."

1280 if

( hdl.

IsNa

() ) {

1283 if

( hdl.

() ) {

1289 if

( num_nucs > 0 && num_prts > 0 ) {

1291 "MIXTURE OF NUCS AND PROTS"

, align);

1306 const

vector< CRef< CSeq_id > >& diag_ids = (*diag_seg)->GetIds();

1307 copy

(diag_ids.begin(), diag_ids.end(), back_inserter(ids));

1317

back_inserter(ids));

1337 id

.GetLabel(&

);

1341 string

seg_string =

"sequence "

","

+ (use_in ?

" in "

" "

) +

1344 ")"

+ (use_in ?

", "

" "

) +

"context "

;

1357 const string

& prefix,

1358 const string

& message)

1363 static const char

kAlignmentTooLong

"the alignment claims to contain residue coordinates that are past the end of the sequence. Either the sequence is too short, or there are extra characters or formatting errors in the alignment"

;

1405 size_t

dim = dendiag.

();

1409 const CSeq_id

& context_id = *(ids[0]);

1410

CDense_diag::TStarts::const_iterator starts_iter =

1413 for

(

size_t id

= 0;

id

< dim; ++id ) {

1417 const CSeq_id

& seq_id = *(ids[id]);

1420 if

( start >= bslen ) {

1425 if

( start +

len

> bslen ) {

1437 int

dim = denseg.GetDim();

1438 size_t

numseg = denseg.GetNumseg();

1444 bool minus

false

;

1445 const CSeq_id

& id_context = *ids[0];

1447 if

(numseg > lens.size()) {

1448

numseg = lens.size();

1451 for

(

unsigned id

= 0;

id

< ids.size(); ++id) {

1453 minus

= denseg.IsSetStrands() &&

1456 for

(

unsigned

seg = 0; seg < numseg; ++seg) {

1458 id

+ (

minus

? numseg - seg - 1 : seg) * dim;

1459 if

(curr_index >= starts.size()) {

1463 if

( starts[curr_index] == -1 ) {

1466 size_t

lens_index =

minus

? numseg - seg - 1 : seg;

1469 if

( starts[curr_index] + lens[lens_index] > bslen ) {

1474 size_t

next_index = curr_index;

1476 for

( next_seg = seg + 1; next_seg < numseg; ++next_seg ) {

1478 id

+ (

minus

? numseg - next_seg - 1 : next_seg) * dim;

1479 if

( starts[next_index] != -1 ) {

1483 if

( next_seg == numseg || next_index == curr_index ) {

1489 if

( starts[curr_index] + (

TSignedSeqPos

)lens[lens_index] !=

1490

starts[next_index] ) {

1492 "Start/Length: There is a problem with "

1493 s_DescribeSegment

(*(ids[

id

]), id_context, seg + 1, starts[curr_index],

true

) +

1494 ": the segment is too long or short or the next segment has an incorrect start position"

, align);

1505 static Uchar

bits[] = { 0x80, 0x40, 0x20, 0x10, 0x08, 0x04, 0x02, 0x01 };

1507 if

(!packed.IsSetDim() || !packed.IsSetIds() || !packed.IsSetPresent() || !packed.IsSetNumseg()) {

1511 size_t

dim = packed.GetDim();

1512 size_t

numseg = packed.GetNumseg();

1515

CPacked_seg::TIds::const_iterator id_it = packed.GetIds().begin();

1516 const CSeq_id

& id_context = **id_it;

1518 for

(

size_t id

= 0;

id

< dim && id_it != packed.GetIds().end(); ++id, ++id_it ) {

1522

(*id_it)->GetLabel (&

);

1523 TSeqPos

seg_start = packed.GetStarts()[id];

1527 for

(

size_t

seg = 0; seg < numseg; ++seg ) {

1528 size_t i

id

+ seg * dim;

1529 if

(

/8 < present.size() && (present[

/ 8] & bits[

% 8]) ) {

1530

seg_start += packed.GetLens()[seg];

1542 const TStd

& std_segs,

1548 const CSeq_id

& id_context = *(stdseg.

GetLoc

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

1556 id

->Assign(*(stdseg.