RetroSearch Browse

().

()) {

89 bool

acceptableDiff =

false

;

92

CUser_object::TData::const_iterator iter1 = aop1->

().begin();

93

CUser_object::TData::const_iterator iter2 = aop2->

().begin();

94 for

(; iter1 != aop1->

().end() && iter2 != aop2->

().end(); ++iter1, ++iter2) {

105 if

(fld1 ==

"FeatureListType"

&& fld2 ==

"FeatureListType"

) {

114 if

(featlisttype1 ==

"Complete Genome"

&& featlisttype2 ==

"Partial Genome"

) {

115

acceptableDiff =

true

;

118 if

(featlisttype1 ==

"Partial Genome"

&& featlisttype2 ==

"Complete Genome"

) {

119

acceptableDiff =

true

;

123

}

else if

(fld1 != fld2) {

129 return

acceptableDiff;

142 for

(

const auto

& se_list_it : seqset.

()) {

148

}

else if

(se.

IsSeq

()) {

157 for

(; seqit; ++seqit) {

161

}

else if

(seqit->IsNa()) {

174

[

this

, &seqset]() {

175

PostErr(eDiag_Warning, eErr_SEQ_PKG_BioseqSetClassNotSet,

176 "Bioseq_set class not set"

, seqset);

180 "Bioseq_set class not set"

, seqset);

211 "Set class should not be conset"

, seqset);

221 if

(nuccnt == 0 && protcnt == 0) {

223 "No Bioseqs in this set"

, seqset);

230 bool

not_all_autodef =

false

;

231 bool

not_same_autodef =

false

;

232 bool

has_any_autodef =

false

;

243 bool

has_autodef =

false

;

249 const CBioseq

& seq = sub->GetSeq();

254

has_any_autodef =

true

;

255 if

(! first_autodef) {

256

first_autodef = aduo;

257

}

else if

(! first_autodef->

(*aduo)) {

258 if

(! second_autodef) {

259

second_autodef = aduo;

262

not_same_autodef =

true

;

264

}

else if

(! second_autodef->

(*aduo)) {

265

not_same_autodef =

true

;

269

has_autodef =

false

;

274

}

else if

(se->IsSeq()) {

275 const CBioseq

& seq = se->GetSeq();

280

has_any_autodef =

true

;

281 if

(! first_autodef) {

282

first_autodef = aduo;

283

}

else if

(! first_autodef->

(*aduo)) {

284 if

(! second_autodef) {

285

second_autodef = aduo;

288

not_same_autodef =

true

;

290

}

else if

(! second_autodef->

(*aduo)) {

291

not_same_autodef =

true

;

295

has_autodef =

false

;

300

not_all_autodef =

true

;

303 if

(not_all_autodef && has_any_autodef) {

305 "Not all pop/phy/mut/eco set components have an autodef user object"

308 if

(not_same_autodef && has_any_autodef) {

310 "Inconsistent autodef user objects in pop/phy/mut/eco set"

319 bool

suppressMissingSetTitle =

false

;

321 if

(has_any_autodef && (! not_all_autodef) && (! not_same_autodef)) {

322

suppressMissingSetTitle =

true

;

341 for

(

const auto

& annot_it : seqset.

GetAnnot

()) {

349

[

this

, &seqset]() { x_ValidateSetDescriptors(seqset

); });

415 if

(cds_seq == contig) {

434 "No nucleotides in nuc-prot set"

, seqset);

435

}

else if

(nuccnt > 1 && segcnt != 1) {

437 "Multiple unsegmented nucleotides in nuc-prot set"

, seqset);

441 "No proteins in nuc-prot set"

, seqset);

444 int

prot_biosource = 0;

447 bool

has_Primary =

false

;

448 bool

has_TPA =

false

;

450

sequence::CDeflineGenerator defline_generator;

453 for

(

const auto

& se_list_it : seqset.

()) {

458 bool

hasMetaGenomeSource =

false

;

460 if

(closest_biosource) {

461 const CBioSource

& src = closest_biosource->GetSource();

464 const COrgMod

& omd = *omd_itr;

466

hasMetaGenomeSource =

true

;

474 for

(

const auto

& it : seq.

().

()) {

481 "DBLink user object should not be on a Bioseq"

, seq);

493 "Nucleotide bioseq should be product of mRNA " 494 "feature on contig, but is not"

498 for

(

const auto

& id_it : seq.

GetId

()) {

500 if

(

id

.IsOther() &&

id

.GetOther().IsSetAccession()) {

505

}

else if

(

id

.IsGenbank() ||

id

.IsEmbl() ||

id

.IsDdbj()) {

507

}

else if

(

id

.IsTpg() ||

id

.IsTpe() ||

id

.IsTpd()) {

512

}

else if

(seq.

IsAa

()) {

516 "Protein bioseq should be product of CDS " 517 "feature on contig, but is not"

521 for

(

const auto

& id_it : seq.

GetId

()) {

523 if

(

id

.IsGenbank() ||

id

.IsEmbl() ||

id

.IsDdbj()) {

525

}

else if

(

id

.IsTpg() ||

id

.IsTpe() ||

id

.IsTpd()) {

532 for

(

const auto

& desc : seq.

().

()) {

533 if

(desc->IsSource()) {

536 if

(desc->IsTitle()) {

537

instantiated = desc->GetTitle();

544 string

generated = defline_generator.GenerateDefline(seq, *

m_Scope

, sequence::CDeflineGenerator::fIgnoreExisting);

546

generated = defline_generator.GenerateDefline(seq, *

m_Scope

, sequence::CDeflineGenerator::fIgnoreExisting | sequence::CDeflineGenerator::fAllProteinNames);

548

instantiated.erase(0, 11);

550

instantiated.erase(0, 12);

552

instantiated.erase(0, 21);

555

generated.erase(0, 11);

557

generated.erase(0, 12);

559

generated.erase(0, 21);

561

generated.erase(0, 21);

568 if

(hasMetaGenomeSource &&

NStr::EqualNocase

(

"MAG: "

+ instantiated, generated)) {

570

}

else if

(hasMetaGenomeSource &&

NStr::EqualNocase

(

"MAG "

+ instantiated, generated)) {

574 "Instantiated protein title does not match automatically " 575 "generated title"

, seq);

591 const string

& set_class = tv->

FindName

(

set

.GetClass(),

true

);

594 "Nuc-prot Bioseq-set contains wrong Bioseq-set, " 595 "its class is \""

+ set_class +

"\"."

set

);

600 if

(prot_biosource > 1) {

603 " proteins with a BioSource descriptor"

, seqset);

604

}

else if

(prot_biosource > 0) {

606 "Nuc-prot set has 1 protein with a BioSource descriptor"

, seqset);

609 bool

has_source =

false

;

610 bool

has_title =

false

;

611 bool

has_refgenetracking =

false

;

613 for

(

const auto

& desc : seqset.

().

()) {

615

&& desc->GetSource().IsSetOrg()

616

&& desc->GetSource().GetOrg().IsSetTaxname()

617

&& !

NStr::IsBlank

(desc->GetSource().GetOrg().GetTaxname())) {

619

}

else if

(desc->IsTitle()) {

621

}

else if

(desc->IsUser() && desc->GetUser().IsRefGeneTracking()) {

622

has_refgenetracking =

true

;

637 "Nuc-prot set does not contain expected BioSource descriptor"

, seqset);

643 "Nuc-prot set should not have title descriptor"

, seqset);

646 if

(has_refgenetracking && (! is_nm)) {

648 "Nuc-prot set should not have RefGeneTracking user object"

, seqset);

651 if

(has_Primary && has_TPA) {

653 "Nuc-prot set should not have a mixture of Primary and TPA accession types"

, seqset);

669 const CMolInfo

* mol_info =

nullptr

;

671 for

(; miit; ++miit) {

677

}

else if

(mol_info->

GetBiomol

() != miit->GetBiomol()) {

685 "Pop/phy/mut/eco set contains inconsistent moltype"

, seqset);

697 "No segmented Bioseq in segset"

, seqset);

704 for

(

const auto

& se_list_it : seqset.

()) {

711

mol = seq_inst.

GetMol

();

715 "Segmented set contains mixture of nucleotides" 716 " and proteins"

, seqset);

720

}

else if

(se.

IsSet

()) {

723 if

(

set

.IsSetClass() &&

727 const string

& set_class_str =

731 "Segmented set contains wrong Bioseq-set, " 732 "its class is \""

+ set_class_str +

"\"."

750 for

(

const auto

& se_list_it : seqset.

()) {

755

mol = seq_inst.

GetMol

();

757

seq_inst_mol = seq_inst.

GetMol

();

761 "Parts set contains mixture of nucleotides " 762 "and proteins"

, seqset);

766

}

else if

(se.

IsSet

()) {

770 const string

& set_class_str =

774 "Parts set contains unwanted Bioseq-set, " 775 "its class is \""

+ set_class_str +

"\"."

790 bool

has_title =

false

;

793 for

(

const auto

& desc : seqset.

().

()) {

794 if

(desc->IsTitle()) {

799 "Only Pop/Phy/Mut/Eco sets should have titles"

803 "Only Pop/Phy/Mut/Eco sets should have titles"

, seqset);

813 if

(! suppressMissingSetTitle) {

815 "Pop/Phy/Mut/Eco set does not have title"

, seqset);

833 "Pop/Phy/Mut/Eco set has no components"

, seqset);

834

}

else if

(seqset.

().size() == 1) {

835 bool

has_alignment =

false

;

837 while

(annot_it && ! has_alignment) {

839

has_alignment =

true

;

843 if

(! has_alignment) {

845 "Pop/Phy/Mut/Eco set has only one component and no alignments"

, seqset);

858 bool

has_title =

false

;

860 for

(

const auto

& desc : seq.

().

()) {

861 if

(desc->IsTitle()) {

869 "Nucleotide component of pop/phy/mut/eco/wgs set is missing its title"

, seq);

884

class_name =

"Pop set"

;

887

class_name =

"Mut set"

;

890

class_name =

"Genbank set"

;

895

class_name =

"Phy/eco/wgs set"

;

898

class_name =

"GenProd set"

;

901

class_name =

"Small genome set"

;

904

class_name =

"Nuc-prot set"

;

912 for

(

const auto

& desc : seqset.

().

()) {

913 if

(desc->IsMolinfo()) {

915

class_name +

" has MolInfo on set"

, seqset);

925 static const string

sp =

" sp. "

;

929 "RefSeq record should not be a Pop-set"

, seqset);

933 string

first_taxname;

934 bool

is_first =

true

;

935 for

(; seqit; ++seqit) {

947 if

(

->GetData().GetBiosrc().IsSetOrg() &&

->GetData().GetBiosrc().GetOrg().IsSetTaxname()) {

948

taxname =

->GetData().GetBiosrc().GetOrg().GetTaxname();

953

first_taxname = taxname;

967 size_t len

= pos + sp.length();

975 size_t

comp_len =

min

(taxname.length(), first_taxname.length());

981 "Population set contains inconsistent organism names."

, seqset);

997 bool

id_no_good =

false

;

1004 "Seq-annot packaged directly on genomic product set"

1008

CBioseq_set::TSeq_set::const_iterator se_list_it =

1011 if

(! (**se_list_it).IsSeq()) {

1015 const CBioseq

& seq = (*se_list_it)->GetSeq();

1027

id_type =

id

.Which();

1030

}

catch

(

const

std::exception&) {

1039 if

(loc_label.empty()) {

1045 "Product of mRNA feature ("

+ loc_label +

1046 ") not packaged in genomic product set"

1053 "Product of mRNA feature (?) not packaged in " 1054 "genomic product set"

1066 for

(

const auto

& se : seqset.

()) {

1068 if

(! se->GetSet().IsSetClass()

1074 "Nested sets within Pop/Phy/Mut/Eco/Wgs set"

1086 for

(

const auto

& it : seqset.

().