RetroSearch Browse

.empty() ? sv : fbp->

;

593 return

std::get<FeatBlk*>(

mData

);

602 for

(

auto

& dbp : dbl) {

619 for

(

char

c :

value

)

642 #ifdef BIOSEQ_FIND_METHOD 644

bsp = BioseqFind(sip);

646 return

(bsp->length);

655 if

(

id

.IsGenbank() ||

id

.IsEmbl() ||

id

.IsDdbj() ||

id

.IsTpg() ||

656 id

.IsTpe() ||

id

.IsTpd())

657

text_id =

id

.GetTextseq_Id();

662 for

(use_indx = 0; use_indx < pp->

indx

; use_indx++) {

664 if

(text_id_acc == e->acnum &&

665

(pp->

accver

false

|| e->vernum == text_id_ver))

669 if

(use_indx >= pp->

indx

) {

674 if

(

static_cast<size_t>

(-1))

675 return static_cast<Int4>

(

);

680 if

(pp->

accver

false

|| text_id_ver < 0)

681 msg

(

"Location points to outside entry {}"

, text_id_acc);

683 msg

(

"Location points to outside entry {}.{}"

, text_id_acc, text_id_ver);

688 if

(! pp->

buf

->empty()) {

689 string msg

(

"Feature location references an interval on another record : {}"

, *pp->

buf

);

723

qual.

GetQual

() !=

"db_xref"

)

738

line = line.substr(4);

740 size_t

colon = line.find(

':'

);

741 if

(colon == string::npos) {

743 "Badly formatted /db_xref qualifier: \"{}\". Qualifier dropped."

);

747 string

tail = line.substr(colon + 1);

748

line = line.substr(0, colon);

752 "/db_xref type \"{}\" is obsolete."

, line);

764 buf

"UniProt/Swiss-Prot"

;

766 buf

"UniProt/TrEMBL"

;

773 string buf

(

"UniProtKB"

);

774 buf

+= line.substr(7);

779 const Char

* strid =

nullptr

;

782 const Char

* p = tail.c_str();

784 for

(strid = p; *p >=

'0'

&& *p <=

'9'

;)

786 if

(*p ==

'\0'

&& *strid !=

'0'

) {

793 for

(strid = p; *p >=

'0'

&& *p <=

'9'

;)

797 "/db_xref qualifier \"{}\" is supposed to be a string, but its value consists of digits only."

);

798 if

(*strid !=

'0'

) {

809 "/db_xref qual should have numeric value greater than 0: \"{}\". Qualifier dropped."

);

814 for

(; *

'0'

&& *

'9'

;)

818 "/db_xref qualifier \"{}\" is supposed to be a numeric identifier, but its value includes alphabetic characters. Qualifier dropped."

);

821 if

'\0'

|| q != p)

828 if

(*p !=

'e'

&& *p !=

'g'

&& *p !=

'd'

) {

830 "Badly formatted /db_xref qual \"PID\": \"{}\". Qualifier dropped."

);

834 const Char

* q = p + 1;

839 for

(

= q; *

'0'

&& *

'9'

;)

841 if

(*q ==

'\0'

|| *

'\0'

) {

843 "/db_xref qual \"PID\" should contain numeric value greater than 0: \"{}\". Qualifier dropped."

);

849 "Unknown data base name /db_xref = \"{}\". Qualifier dropped."

);

859 tag

->SetTag().SetStr(strid);

861 tag

->SetTag().SetId(intid);

892 for

(CSeq_feat::TQual::iterator qual = feat.

().begin(); qual != feat.

().end();) {

893 if

(! (*qual)->IsSetQual() || (*qual)->GetQual() !=

"db_xref"

) {

904

db_refs.push_back(dbtag);

909

qual = feat.

().erase(qual);

955 if

(cur_loc->

IsInt

())

984 return

make_unique<string>(

);

986 for

(ptr =

; *ptr !=

' '

&& *ptr !=

'\0'

;)

995 while

(*eptr ==

' '

|| *eptr ==

')'

)

999 return

make_unique<string>(ptr, eptr);

1071 const CSeq_id

* cur_id =

nullptr

;

1073 switch

(cur_loc->

Which

()) {

1077

cur_id = cur_loc->

();

1083

cur_id = cur_loc->

();

1089

cur_id = cur_loc->

();

1097

cur_id = cur_loc->

();

1102

cur_id = cur_loc->

();

1112 if

(! accession || ! cur_id)

1129 else if

(strand != cur_loc->

GetStrand

())

1147 while

((q =

.find(

"aa"

)) != string::npos) {

1173

dbp = temp_xml_chain.begin();

1174

dbp_end = temp_xml_chain.end();

1177

dbp = chain.begin();

1178

dbp_end = chain.end();

1183 for

(; dbp != dbp_end; ++dbp) {

1184 auto

& ref_blk = *dbp;

1189 if

(pubdesc.

())

1193

feat->

().SetPub(*pubdesc);

1195

unique_ptr<string> ploc;

1201 auto i

= q->find(

'('

);

1202 if

(

!= string::npos)

1206 if

(ploc->find(

';'

) != string::npos) {

1207 string

s =

"join("

;

1214 for

(p = ref_blk.mBuf.ptr + col_data; *p !=

'\0'

&& *p !=

'('

;)

1216

ploc =

CheckLocStr

(

string

(p, ref_blk.mBuf.ptr + ref_blk.mBuf.len - p).c_str());

1218 for

(

const auto

& subdbp : ref_blk.GetSubBlocks()) {

1222 for

(p = subdbp.mBuf.ptr; *p !=

'\0'

isdigit

(*p) == 0;)

1225 string

s =

"join("

;

1228

ploc = make_unique<string>(s);

1230

ploc = make_unique<string>(p);

1234 if

(! ploc || ploc->empty()) {

1258

feats.push_back(feat);

1263 "Mixed strands in SeqLoc: {}"

);

1265

feats.push_back(feat);

1295

dbp = temp_xml_chain.begin();

1296

dbp_end = temp_xml_chain.end();

1299

dbp = chain.begin();

1300

dbp_end = chain.end();

1306 for

(; dbp != dbp_end; ++dbp) {

1307 auto

& ref_blk = *dbp;

1312 if

(pubdesc.

() || ! pubdesc->IsSetPub())

1319

imp_feat.

SetKey

(

"Site-ref"

);

1320

imp_feat.

SetLoc

(

"sites"

);

1327

pub->

SetEquiv

(pubdesc->SetPub());

1329

feat->

SetCit

().SetPub().push_back(pub);

1331 if

(pubdesc->IsSetComment())

1338

feats.push_back(feat);

1353 return "unknown location"

;

1375 for

(q = p + 4; *q ==

' '

;)

1378 for

(pars = 0, p = q; *p !=

'\0'

; p++) {

1379 if

(*p ==

','

&& pars == 0)

1383 else if

(*p ==

')'

) {

1393 const char

* loc_str = loc_.c_str();

1399 if

(ret.

()) {

1402 "Invalid position element for an /anticodon qualifier : \"{}\" : qualifier dropped : feature location \"{}\"."

, loc_str, loc.empty() ?

"unknown"

s : loc);

1411 "tRNA feature at \"{}\" has anticodon with location spanning four bases: \"{}\". Cannot generate corresponding codon value from the DNA sequence."

, loc.empty() ?

"unknown"

s : loc, loc_str);

1414 "tRNA feature at \"{}\" has anticodon of an unusual size: \"{}\". Cannot generate corresponding codon value from the DNA sequence."

, loc.empty() ?

"unknown"

s : loc, loc_str);

1423 if

(xrange != anticodon_range) {

1426 "Anticodon location \"{}\" does not fall within tRNA feature at \"{}\"."

, loc_str, loc.empty() ?

"unknown"

s : loc);

1454 len

= comment.size();

1456 if

(

> 15 &&

< 20) {

1457 if

(

StringEquNI

(comment.c_str() +

- 15,

"S ribosomal RNA"

, 15)) {

1461

}

else if

(

> 6 &&

< 20) {

1469 if

(qval_str.empty())

1473 for

(p = qval; p; p += 13) {

1480 for

(p = qval; p; p = qval +

) {

1488 len

= p - qval + 13;

1498

s.append(

" ribosomal RNA"

);

1505 for

(p = qval, q = p; q; q = p + 13) {

1518 if

(p && p > qval && p[15] ==

'\0'

) {

1520 if

(*p >=

'0'

&& *p <=

'9'

)

1528 if

(p == qval || (p[9] !=

' '

&& p[9] !=

'\0'

)) {

1536 len

= p - qval + 14;

1567

rna_ref.

().SetName(qval);

1577 if

(acp->

intaa

== ch)

1580 return

(acp->

intaa

);

1591 for

(tap =

taa

; tap->

name

; tap++)

1601 return

(acp->

intaa

);

1637 if

(product.length() < 7)

1640 bool

digits =

false

;

1642 for

(p = prod; *p !=

'\0'

; p++) {

1643 if

(*p >=

'a'

&& *p <=

'z'

)

1645 else if

((*p < 'A' || *p >

'Z'

) && *p !=

'('

&& *p !=

')'

) {

1646 if

(*p >=

'0'

&& *p <=

'9'

)

1664 for

(p = end; *p !=

'\0'

; p++)

1665 if

(*p ==

'('

|| *p ==

')'

)

1669 if

(start == prod && *end ==

'\0'

) {

1680 for

(p = end; *p ==

' '

|| *p ==

')'

|| *p ==

'('

;)

1687 while

(*p >=

'A'

&& *p <=

'Z'

)

1694 while

(*p ==

' '

|| *p ==

')'

|| *p ==

'('

)

1696 for

(q = p; *p >=

'A'

&& *p <=

'Z'

;)

1701 if

(q[1] ==

'\0'

) {

1702 while

(*p ==

' '

|| *p ==

')'

|| *p ==

'('

)

1704 for

(q = p; *p >=

'A'

&& *p <=

'Z'

;)

1714 while

(*p ==

' '

|| *p ==

'('

|| *p ==

')'

)

1720 for

(p = start - 1; *p ==

' '

|| *p ==

')'

|| *p ==

'('

; p--)

1724 if

(p > prod && p[1] ==

')'

) {

1725 for

(p--; *p !=

'('

; p--)

1729 for

(p--; *p ==

' '

|| *p ==

'('

|| *p ==

'('

; p--)

1735 for

(q = p++; *q >=

'A'

&& *q <=

'Z'

; q--)

1738 if

(*q < 'A' || *q >

'Z'

)

1767 if

first

&& ! second && ! third && ! fourth &&

remove

&& ! digits)

1778

comment +=

"; fMet"

;

1794 if

(comment.empty())

1798 for

(p = comm; *p !=

'\0'

; p++) {

1799 if

(*p >=

'a'

&& *p <=

'z'

)

1801 else if

(*p < 'A' || *p >

'Z'

)

1806 if

(

StringEquN

(comm,

"CODON RECOGNIZED "

, 17)) {

1809 if

(q &&

(q + 1,

"PUTATIVE"

))

1818 if

(

StringEquN

(comm,

"PUTATIVE "

, 9) && comm[10] ==

' '

1819

comm[14] ==

' '

StringEquN

(&comm[15],

"TRNA"

, 4)) {

1827 for

(q = comm, p = q; p;) {

1855

optional<string> qval;

1876

feat.

().SetRna(*rna_ref);

1890

rna_gen->SetClass(*p);

1896

rna_qual->

(

"tag_peptide"

);

1900

rna_quals->

Set

().push_back(rna_qual);

1903

rna_gen->SetQuals(*rna_quals);

1912 if

(p && ! p->empty()) {

1929 const Char

* c_q =

nullptr

;

1930 for

(;; c_p += 5, c_q = c_p) {

1936 const Char

* c_r =

nullptr

;

1937 for

(c_p = feat.

GetComment

().c_str();; c_p += 4, c_r = c_p) {

1944

c_p = (c_q > c_r) ? c_q : c_r;

1951 while

(*c_p ==

' '

|| *c_p ==

'\t'

|| *c_p ==

','

|| *c_p ==

';'

)

1954 if

(*c_p ==

'\0'

) {

1962 if

(qval->length() > 511) {

1964

qval->back() =

'>'

;

1968 if

(rna_gen.

())

1971

rna_gen->SetProduct(*qval);

1973

rna_ref->

().SetName(*qval);

1983

rna_ref->

().SetGen(*rna_gen);

2004

trnaa->SetAnticodon(*anticodon);

2005

rna_ref->

().SetTRNA(*trnaa);

2013 if

(! qval2.empty()) {

2036 if

(trnaa.

()) {

2037 if

(trnap.

()) {

2039

rna_ref->

().SetTRNA(*trnac);

2045

rna_ref->

().SetTRNA(*trnap);

2049

rna_ref->

().SetTRNA(*trnac);

2053

trnap->SetCodon().assign(trnac->GetCodon().begin(), trnac->GetCodon().end());

2072

trnac->SetAnticodon(trnaa->SetAnticodon());

2073

trnaa->ResetAnticodon();

2076

trnac->SetCodon().assign(trnaa->GetCodon().begin(), trnaa->GetCodon().end());

2079

rna_ref->

().SetTRNA(*trnac);

2116

feat.

().SetImp(*imp_feat);

2123 for

(COrg_ref::TDb::iterator db = bio.

SetOrg

().SetDb().begin(); db != bio.

SetOrg

().SetDb().end(); ++db) {

2124 if

(! (*db)->CanGetDb())

2127

COrg_ref::TDb::iterator tdb = db;

2128 for

(++tdb; tdb != bio.

SetOrg

().SetDb().end(); ++tdb) {

2129 if

(! (*tdb)->IsSetDb())

2132 if

((*db)->GetDb() < (*tdb)->GetDb())

2135 if

((*db)->GetDb() == (*tdb)->GetDb()) {

2137 const CObject_id

& tdb_id = (*tdb)->GetTag();

2146 if

(! db_id.

IsStr

() && ! tdb_id.

IsStr

() &&

2157 for

(COrgName::TMod::iterator

= rmod.begin();

!= rmod.end(); ++

) {

2158

COrgName::TMod::iterator tmod =

;

2159 for

(++tmod; tmod != rmod.end(); ++tmod) {

2160 if

((*mod)->GetSubtype() < (*tmod)->GetSubtype())

2163 if

((*mod)->GetSubtype() == (*tmod)->GetSubtype() &&

2164

(*mod)->GetSubname() <= (*tmod)->GetSubname())

2177 for

(CBioSource::TSubtype::iterator sub = rsub.begin(); sub != rsub.end(); ++sub) {

2178

CBioSource::TSubtype::iterator tsub = sub;

2179 for

(++tsub; tsub != rsub.end(); ++tsub) {

2180 if

((*sub)->GetSubtype() < (*tsub)->GetSubtype())

2183 if

((*sub)->GetSubtype() == (*tsub)->GetSubtype() &&

2184

(*sub)->GetName() <= (*tsub)->GetName())

2196 bool

has_comma =

.find(

','

) != string::npos;

2199

std::replace(

.begin(),

.end(),

','

';'

);

2210 if

(! fbp || fbp->

.empty())

2213

TQualVector::iterator

first

= fbp->

.end();

2216 for

(TQualVector::iterator qual = fbp->

.begin(); qual != fbp->

.end();) {

2217 if

((*qual)->GetQual() !=

"rpt_unit"

) {

2223 "Obsolete /rpt_unit qualifier found on feature \"{}\" at location \"{}\"."

(fbp,

"Unknown"

(fbp,

"unknown"

));

2225 if

((*qual)->GetVal().empty()) {

2226

qual = fbp->

.erase(qual);

2231 len

+= (*qual)->GetVal().size();

2251 const string

= (*first)->GetVal();

2252 if

.begin() ==

'('

&& *

.rbegin() ==

')'

) {

2261

p.append((*first)->GetVal());

2263 for

(TQualVector::iterator qual =

first

; qual != fbp->

.end();) {

2264 if

((*qual)->GetQual() !=

"rpt_unit"

) {

2270

p.append((*qual)->GetVal());

2271

qual = fbp->

.erase(qual);

2274

(*first)->SetVal(p);

2287 if

(! fbp || fbp->

.empty())

2297 for

(TQualVector::iterator qual = fbp->

.begin(); qual != fbp->

.end();) {

2298 const string

& qual_str = (*qual)->IsSetQual() ? (*qual)->GetQual() :

""

;

2299 const string

& val_str = (*qual)->IsSetVal() ? (*qual)->GetVal() :

""

;

2300 if

(qual_str ==

"experiment"

) {

2301 if

(val_str ==

"experimental evidence, no additional details recorded"

) {

2303

qual = fbp->

.erase(qual);

2311 if

(qual_str ==

"inference"

) {

2312 if

(val_str ==

"non-experimental evidence, no additional details recorded"

) {

2314

qual = fbp->

.erase(qual);

2322 if

(qual_str !=

"evidence"

) {

2333 "Illegal value \"{}\" for /evidence qualifier on the \"{}\" feature at \"{}\". Qualifier dropped."

, val_str.empty() ?

"Unknown"

s : val_str,

(fbp,

"Unknown"

(fbp,

"unknown location"

));

2336

qual = fbp->

.erase(qual);

2339 if

(evi_exp + evi_not > 0 && exp_good + exp_bad + inf_good + inf_bad > 0) {

2341 "Old /evidence and new /experiment or /inference qualifiers both exist on the \"{}\" feature at \"{}\". This is currently unsupported."

(fbp,

"Unknown"

(fbp,

"unknown location"

));

2345 if

(evi_exp + exp_good > 0 && evi_not + inf_good > 0) {

2347 "The special \"no additional details recorded\" values for both /experiment and /inference exist on the \"{}\" feature at \"{}\". This is currently unsupported."

(fbp,

"Unknown"

(fbp,

"unknown location"

));

2351 if

((exp_good > 0 && exp_bad > 0) || (inf_good > 0 && inf_bad > 0)) {

2353 "The special \"no additional details recorded\" value for /experiment or /inference exists in conjunction with other /experiment or /inference qualifiers on the \"{}\" feature at \"{}\". This is currently unsupported."

(fbp,

"Unknown"

(fbp,

"unknown location"

));

2357 if

(exp_good + evi_exp > 0)

2359 else if

(inf_good + evi_not > 0)

2384 bool

locmap =

false

;

2392

pp->

buf

= fbp->

" : "

+ loc;

2398 if

(pp->

debug

false

) {

2400 "{}|{}| range check detects problems"

, fbp->

, loc);

2405 "{}|{}| range check detects problems"

, fbp->

, loc);

2409 if

(! fbp->

.empty()) {

2414 if

(loc.find(

"order"

) != string_view::npos)

2417 if

(! fbp->

.empty()) {

2422 if

(! fbp->

.empty())

2425 if

(! fbp->

.empty())

2428 if

(! fbp->

.empty()) {

2430 if

(fbp->

== *

)

2438

exc_text +=

", trans-splicing"

;

2453 if

(! fbp->

.empty()) {

2456 if

(! comment->empty()) {

2464 if

(fbp->

.find(

"source"

) == string::npos)

2467 for

(

const auto

& cur : fbp->

) {

2468 const string

& qual_str = cur->GetQual();

2469 if

(qual_str ==

"pseudogene"

)

2473 if

(qual_str ==

"translation"

&& (! cur->IsSetVal() || cur->GetVal().empty()))

2476 if

(! qual_str.empty())

2477

feat->

().push_back(cur);

2499 for

(TQualVector::iterator q = fbp->

.begin(); q != fbp->

.end(); ++q) {

2500 if

((*q)->GetQual() ==

"gene"

2501

(! qamode && (*q)->GetQual() ==

"product"

))

2504

TQualVector::iterator tq = q;

2505 for

(++tq; tq != fbp->

.end(); ++tq) {

2506 const string

& q_qual = (*q)->GetQual();

2507 const string

& tq_qual = (*tq)->GetQual();

2509 if

(! tq_qual.empty()) {

2510 if

(q_qual ==

"gene"

)

2519 const string

q_val = (*q)->GetVal();

2520 const string

tq_val = (*tq)->GetVal();

2525 if

(! tq_val.empty()) {

2526 if

(q_val[0] >=

'0'

&& q_val[0] <=

'9'

2527

tq_val[0] >=

'0'

&& tq_val[0] <=

'9'

) {

2530

}

else if

(q_val <= tq_val)

2544 bool

found =

false

;

2546 for

(

const auto

& gbqp1 : qual1) {

2548 for

(

const auto

& gbqp2 : qual2) {

2549 const Char

* qual_a = gbqp1->IsSetQual() ? gbqp1->GetQual().c_str() :

nullptr

;

2550 const Char

* qual_b = gbqp2->IsSetQual() ? gbqp2->GetQual().c_str() :

nullptr

;

2552 const Char

* val_a = gbqp1->IsSetVal() ? gbqp1->GetVal().c_str() :

nullptr

;

2553 const Char

* val_b = gbqp2->IsSetVal() ? gbqp2->GetVal().c_str() :

nullptr

;

2573 if

(! fbp1 && ! fbp2)

2575 if

(! fbp1 || ! fbp2 ||

2594 if

|| *

'\0'

)

2597 for

(p =

; *p >=

'0'

&& *p <=

'9'

;)

2600 if

(p ==

|| p[0] !=

'.'

|| p[1] !=

'.'

)

2604 for

(p += 2, q = p; *q >=

'0'

&& *q <=

'9'

;)

2606 if

(q == p || *q !=

'\0'

)

2610 if

(i1 == 0 || i1 > i2 || i2 > (

Int4

)length)

2618 if

(! fbp || fbp->

.empty())

2621 for

(TQualVector::iterator cur = fbp->

.begin(); cur != fbp->

.end();) {

2622 if

(! (*cur)->IsSetQual() || ! (*cur)->IsSetVal()) {

2627 const string

& qual_str = (*cur)->GetQual();

2628 const string

& val_str = (*cur)->GetVal();

2636 if

(_loc.size() > 20) {

2641 "/rpt_unit_range qualifier \"{}\" on feature \"{}\" at location \"{}\" is not a valid basepair range. Qualifier dropped."

, val_str.empty() ?

"(EMPTY)"

s : val_str,

(fbp,

"Unknown"

), _loc);

2643

cur = fbp->

.erase(cur);

2650 auto

dbp = dbl.begin();

2651 if

(dbp == dbl.end() ||

(dbp) == dbl.end())

2654 for

(; dbp != dbl.end(); ++dbp) {

2655 if

(! dbp->hasData())

2658 const FeatBlk

* fbp1 = dbp->GetFeatData();

2660 auto

tdbpprev = dbp;

2661 for

(

auto

tdbp =

(dbp); tdbp != dbl.end();) {

2662 if

(! tdbp->hasData()) {

2663

tdbp = dbl.erase_after(tdbpprev);

2667 const FeatBlk

* fbp2 = tdbp->GetFeatData();

2680 if

(_loc.size() > 20) {

2685 "Duplicated feature \"{}\" at location \"{}\" removed."

(fbp2,

"???"

), _loc);

2688

tdbp = dbl.erase_after(tdbpprev);

2711 for

(

const auto

& dbp : dbl) {

2712 const FeatBlk

* fbp = dbp.GetFeatData();

2721 "Multiple /locus_tag values for \"{}\" feature at \"{}\"."

(fbp,

"Unknown"

(fbp,

"unknown location"

));

2731

isLocusTag(

"locus_tag"

);

2733 for

(

const auto

& dbp : dbl) {

2734 const FeatBlk

* fbp = dbp.GetFeatData();

2737 size_t

olt = std::count_if(fbp->

.begin(), fbp->

.end(), isOldLocusTag);

2738 size_t lt

= std::count_if(fbp->

.begin(), fbp->

.end(), isLocusTag);

2745 "Feature \"{}\" at \"{}\" has an /old_locus_tag qualifier but lacks a /locus_tag qualifier. Entry dropped."

(fbp,

"Unknown"

(fbp,

"unknown location"

));

2748 for

(

const auto

& gbqp1 : fbp->

) {

2749 if

(! gbqp1->IsSetQual() || ! gbqp1->IsSetVal() || ! isLocusTag(gbqp1))

2752 const string

& gbqp1_val = gbqp1->GetVal();

2753 if

(gbqp1_val.empty())

2756 for

(

const auto

& gbqp2 : fbp->

) {

2757 if

(! gbqp2->IsSetQual() || ! gbqp2->IsSetVal())

2760 const string

& gbqp2_val = gbqp2->GetVal();

2765 "Feature \"{}\" at \"{}\" has an /old_locus_tag qualifier with a value that is identical to that of a /locus_tag qualifier: \"{}\". Entry dropped."

(fbp,

"Unknown"

(fbp,

"unknown location"

), gbqp1_val);

2774 for

(TQualVector::const_iterator gbqp1 = fbp->

.begin(); gbqp1 != fbp->

.end(); ++gbqp1) {

2775 if

(! (*gbqp1)->IsSetVal())

2777 const string

& gbqp1_val = (*gbqp1)->GetVal();

2778 if

(isOldLocusTag(*gbqp1) || gbqp1_val.empty())

2781

TQualVector::const_iterator gbqp2 = gbqp1;

2782 for

(++gbqp2; gbqp2 != fbp->

.end(); ++gbqp2) {

2783 const string

& gbqp2_val = (*gbqp2)->GetVal();

2784 if

(isOldLocusTag(*gbqp2) || gbqp2_val.empty())

2789 "Feature \"{}\" at \"{}\" has redundant /old_locus_tag qualifiers. Dropping all but the first."

(fbp,

"Unknown"

(fbp,

"unknown location"

));

2794 if

(gbqp2 != fbp->

.end())

2803 for

(

auto

& dbp : dbl) {

2808 bool

got_pseudo =

false

;

2809 bool

got_pseudogene =

false

;

2811 for

(TQualVector::iterator cur = fbp->

.begin(); cur != fbp->

.end();) {

2812 const string

& qual_str = (*cur)->GetQual();

2813 const string

& val_str = (*cur)->IsSetVal() ? (*cur)->GetVal() :

""

;

2815 if

(qual_str !=

"pseudogene"

) {

2816 if

(! got_pseudo && qual_str ==

"pseudo"

)

2822 if

(got_pseudogene) {

2824 "Dropping a /pseudogene qualifier because multiple /pseudogene qualifiers are present : <{}> : Feature key <{}> : Feature location <{}>."

, val_str.empty() ?

"[empty]"

s : val_str, fbp->

, *fbp->

);

2826

cur = fbp->

.erase(cur);

2830

got_pseudogene =

true

;

2832 if

(val_str.empty()) {

2834 "Dropping a /pseudogene qualifier because its value is empty : Feature key <{}> : Feature location <{}>."

, fbp->

, *fbp->

);

2836

cur = fbp->

.erase(cur);

2846 "Dropping a /pseudogene qualifier because its value is invalid : <{}> : Feature key <{}> : Feature location <{}>."

, val_str, fbp->

, *fbp->

);

2848

cur = fbp->

.erase(cur);

2851 if

(! got_pseudogene || ! got_pseudo)

2855 "A legacy /pseudo qualifier and a /pseudogene qualifier are present on the same feature : Dropping /pseudo : Feature key <{}> : Feature location <{}>."

, fbp->

, *fbp->

);

2863 for

(

auto

& dbp : dbl) {

2864 FeatBlk

* fbp = dbp.GetFeatData();

2871 for

(TQualVector::iterator cur = fbp->

.begin(); cur != fbp->

.end();) {

2872 const string

& qual_str = (*cur)->GetQual();

2873 const string dummy

;

2874 const string

& val_str = (*cur)->IsSetVal() ? (*cur)->SetVal() :

dummy

;

2876 if

(qual_str ==

"compare"

) {

2877 bool

badcom =

true

;

2878 if

(! val_str.empty()) {

2879 const char

* q =

StringChr

(val_str.c_str(),

'.'

);

2880 if

(q && q[1] !=

'\0'

) {

2882 for

(p = q + 1; *p >=

'0'

&& *p <=

'9'

;)

2893 "/compare qualifier value is not a legal Accession.Version : feature \"{}\" at \"{}\" : value \"{}\" : qualifier has been dropped."

, fbp->

, *fbp->

, val_str.empty() ?

"[empty]"

s : val_str);

2894

cur = fbp->

.erase(cur);

2898

}

else if

(qual_str ==

"citation"

)

2904 if

(com_count > 0 || cit_count > 0 ||

2905

(fbp->

"old_sequence"

&& fbp->

"conflict"

))

2909 "Feature \"{}\" at \"{}\" lacks required /citation and/or /compare qualifier : feature has been dropped."

, fbp->

, *fbp->

);

2926 auto

dbp = dbl.begin();

2927 for

(; dbp != dbl.end(); ++dbp) {

2928

fbp = dbp->GetFeatData();

2932 for

(p =

, q = p; *p !=

'\0'

; p++)

2933 if

(*p !=

' '

&& *p !=

'\t'

&& *p !=

'\n'

)

2942 for

(p =

+ 1; *p !=

'\0'

; p++) {

2945 for

(

nullptr

, q = p - 1;; q--) {

2947 if

(*q !=

'_'

&& (*q < '0' || *q >

'9'

) &&

2948

(*q < 'a' || *q >

'z'

) && (*q < 'A' || *q >

'Z'

))

2960 if

(*q !=

'_'

&& (*q < '0' || *q >

'9'

) &&

2961

(*q < 'a' || *q >

'z'

) && (*q < 'A' || *q >

'Z'

)) {

2970

(q[1] ==

'z'

|| q[1] ==

'Z'

) && ibp->

is_tpa

false

)

2984 if

(dbp == dbl.end())

2994 "Feature \"{}\" at \"{}\" on a TSA record cannot point to a non-TSA record."

, fbp->

"empty_location"

);

2995

}

else if

(ibp->

is_tls

) {

2997 "Feature \"{}\" at \"{}\" on a TLS record cannot point to a non-TLS record."

, fbp->

"empty_location"

);

3000 "Feature \"{}\" at \"{}\" on a TPA record cannot point to a non-TPA record."

, fbp->

"empty_location"

);

3009 using

FTAOperonList = list<FTAOperon*>;

3010

FTAOperonList operonList;

3011

FTAOperonList residentList;

3012 bool

success =

true

;

3014 if

(feats.empty()) {

3018 for

(

const auto

& pFeat : feats) {

3019 if

(! pFeat->GetData().IsImp())

3022 const auto

& featLocation = pFeat->GetLocation();

3023 const CImp_feat

& featImp = pFeat->GetData().GetImp();

3027 for

(

const auto

& pQual : pFeat->GetQual()) {

3028 const auto

& qual = *pQual;

3029 if

(! qual.IsSetQual() || qual.GetQual() !=

"operon"

3030

! qual.IsSetVal() || qual.GetVal().empty()) {

3040

operonList.push_back(pLatest);

3042

residentList.push_back(pLatest);

3045 for

(

const auto

& operon : operonList) {

3046 if

(pLatest == operon) {

3049 if

(pLatest->

mOperon

!= operon->mOperon) {

3053 "The operon features at \"{}\" and \"{}\" utilize the same /operon qualifier : \"{}\"."

, operon->LocationStr(), pLatest->

LocationStr

(), pLatest->

mOperon

);

3058 if

(opQualCount > 1) {

3060 "Feature \"{}\" at \"{}\" has more than one operon qualifier."

, pLatest->

mFeatname

, pLatest->

LocationStr

());

3064 if

(opQualCount == 0 && featImp.

IsSetKey

() && featImp.

() ==

"operon"

) {

3071 for

(

const auto

& resident : residentList) {

3072 bool

matched =

false

;

3073 for

(

const auto

& operon : operonList) {

3074 if

(resident->mOperon != operon->mOperon) {

3082 "Feature \"{}\" at \"{}\" with /operon qualifier \"{}\" does not fall within the span of the operon feature at \"{}\"."

, resident->mFeatname, resident->LocationStr(), resident->mOperon, operon->LocationStr());

3088 "/operon qualifier \"{}\" on feature \"{}\" at \"{}\" has a value that does not match any of the /operon qualifiers on operon features."

, resident->mOperon, resident->mFeatname, resident->LocationStr());

3092 for

(

auto

& resident : residentList) {

3095 for

(

auto

& operon : operonList) {

3104 if

(! fbp || fbp->

.empty())

3107 for

(TQualVector::iterator cur = fbp->

.begin(); cur != fbp->

.end(); ++cur) {

3108 const char

* cur_qual = (*cur)->IsSetQual() ? (*cur)->GetQual().c_str() :

nullptr

;

3109 const char

* cur_val = (*cur)->IsSetVal() ? (*cur)->GetVal().c_str() :

nullptr

;

3111

TQualVector::iterator

= cur;

3113 const char

* next_qual = (*next)->IsSetQual() ? (*next)->GetQual().c_str() :

nullptr

;

3114 const char

* next_val = (*next)->IsSetVal() ? (*next)->GetVal().c_str() :

nullptr

;

3122 if

(_loc.size() > 20) {

3128 "Duplicated qualifier \"{}\" in feature \"{}\" at location \"{}\" removed."

, cur_qual ? cur_qual :

"???"

(fbp,

"???"

), _loc);

3142

list<string> linkage_evidence_names;

3145 const char

* gap_type;

3148 Int4

estimated_length;

3171 if

(is_htg >= 0 && is_htg <= 2)

3175 else if

(

"HTGS_PHASE0"

)

3177 else if

(

"HTGS_PHASE1"

)

3179 else if

(

"HTGS_PHASE2"

)

3181 else if

(

"HTGS_PHASE3"

)

3187

finished_gap =

false

;

3188

ibp->

.clear();

3189 for

(; dbp != dbp_end; ++dbp) {

3192 if

(dbp->mType !=

)

3195

linkage_evidence_names.clear();

3196

asn_linkage_evidence.clear();

3198 for

(

const auto

& tdbp : dbp->GetSubBlocks()) {

3201 const FeatBlk

* fbp = tdbp.GetFeatData();

3202 if

(! fbp || fbp->

.empty())

3204 if

(fbp->

"gap"

) {

3205

prev_gap = curr_gap;

3207

}

else if

(fbp->

"assembly_gap"

) {

3208

prev_gap = curr_gap;

3215

gap_type =

nullptr

;

3216

linkage_evidence_names.clear();

3218

asn_linkage_evidence.clear();

3219

estimated_length = -1;

3221 for

(

const auto

& cur : fbp->

) {

3222 if

(! cur->IsSetQual() || ! cur->IsSetVal())

3225 const string

& cur_qual = cur->GetQual();

3226 const string

& cur_val = cur->GetVal();

3228 if

(cur_qual.empty() || cur_val.empty())

3231 if

(cur_qual ==

"estimated_length"

) {

3232 if

(cur_val ==

"unknown"

)

3233

estimated_length = -100;

3235 if

(string::npos == cur_val.find_first_not_of(

"0123456789"

))

3238

}

else if

(cur_qual ==

"gap_type"

)

3239

gap_type = cur_val.c_str();

3240 else if

(cur_qual ==

"linkage_evidence"

) {

3241

linkage_evidence_names.push_back(cur_val);

3248 bool

has_lt =

false

;

3249 if

(q.starts_with(

'<'

)) {

3253 size_t

p = q.find_first_not_of(

"0123456789"

);

3255 if

(p == string_view::npos) {

3259 if

(q.starts_with(

'.'

)) {

3261 if

(has_lt && from != 1)

3263 else if

(q.starts_with(

'.'

)) {

3265 bool

has_gt =

false

;

3266 if

(q.starts_with(

'>'

)) {

3270 if

(string_view::npos == q.find_first_not_of(

"0123456789"

))

3272 if

(has_gt && to != (

int

)ibp->

bases

)

3279 if

(from == 0 || to == 0 || from > to) {

3280 if

(curr_gap == 1) {

3282 "Invalid gap feature location : \"{}\" : all gap features must have a simple X..Y location on the plus strand."

(fbp,

"unknown"

));

3285 "Invalid assembly_gap location : \"{}\"."

(fbp,

"unknown"

));

3293 if

(gap_type && is_htg > -1 &&

3294

(gap_type,

"within scaffold"

) &&

3295

(gap_type,

"repeat within scaffold"

)) {

3297 "assembly_gap has /gap_type of \"{}\", but clone-based HTG records are only expected to have \"within scaffold\" or \"repeat within scaffold\" gaps. assembly_gap feature located at \"{}..{}\"."

, gap_type, from, to);

3300 if

(is_htg == 0 || is_htg == 1) {

3301 for

(

const string

& evidence : linkage_evidence_names) {

3304 "assembly gap has /linkage_evidence of \"{}\", but unoriented and unordered Phase0/Phase1 HTG records are expected to have \"unspecified\" evidence. assembly_gap feature located at \"{}..{}\"."

, evidence, from, to);

3307

}

else if

(is_htg == 2 || is_htg == 3) {

3308 for

(

const string

& evidence : linkage_evidence_names) {

3312 "assembly gap has /linkage_evidence of \"unspecified\", but ordered and oriented HTG records are expected to have some level of linkage for their gaps. assembly_gap feature located at \"{}..{}\"."

, from, to);

3316 if

(is_htg == 3 && ! finished_gap) {

3318 "Finished Phase-3 HTG records are not expected to have any gaps. First assembly_gap feature encountered at \"{}..{}\"."

, from, to);

3319

finished_gap =

true

;

3324 "assembly_gap feature at \"{}..{}\" lacks the required /gap_type qualifier."

, from, to);

3334 "assembly_gap feature at \"{}..{}\" has an invalid gap type : \"{}\"."

, from, to, gap_type);

3338

asn_gap_type = snp->

;

3340 if

(linkage_evidence_names.empty() &&

3341

(

(gap_type,

"within scaffold"

) ||

3342 StringEqu

(gap_type,

"repeat within scaffold"

))) {

3344 "assembly_gap feature at \"{}..{}\" with gap type \"{}\" lacks a /linkage_evidence qualifier."

, from, to, gap_type);

3348 if

(! linkage_evidence_names.empty()) {

3349 if

(gap_type,

"unknown"

) &&

3350

(gap_type,

"within scaffold"

) &&

3351

(gap_type,

"repeat within scaffold"

)) {

3353 "The /linkage_evidence qualifier is not legal for the assembly_gap feature at \"{}..{}\" with /gap_type \"{}\"."

, from, to, gap_type);

3358 for

(

const string

& evidence : linkage_evidence_names) {

3360 if

(evidence == snp->

)

3364 "assembly_gap feature at \"{}..{}\" has an invalid linkage evidence : \"{}\"."

, from, to, evidence);

3370

new_evidence->SetType(snp->

);

3371

asn_linkage_evidence.push_back(new_evidence);

3376 if

(prev_gap + curr_gap == 3) {

3379 msg

(

"Legacy gap feature at \"{}..{}\" co-exists with a new AGP 2.0 assembly_gap feature at \"{}..{}\"."

, from, to, gfp->from, gfp->to);

3381 msg

(

"Legacy gap feature at \"{}..{}\" co-exists with a new AGP 2.0 assembly_gap feature at \"{}..{}\"."

, gfp->from, gfp->to, from, to);

3387 if

(estimated_length == -1)

3390 "The gap feature at \"{}..{}\" lacks the required /estimated_length qualifier."

, from, to);

3392

}

else if

(estimated_length == 0) {

3394 "Gap feature at \"{}..{}\" has an illegal /estimated_length qualifier : \"{}\" : should be \"unknown\" or an integer."

, from, to,

""

);

3397

}

else if

(estimated_length == -100) {

3398 if

(is_htg >= 0 && to - from != 99) {

3400 "Gap feature at \"{}..{}\" has /estimated_length \"unknown\" but the gap size is not 100 bases."

, from, to);

3402

}

else if

(estimated_length != to - from + 1) {

3411 "Gap feature at \"{}..{}\" has a size that does not match the /estimated_length : {}."

, from, to, estimated_length);

3414 for

(gfp = ibp->

.begin(); gfp != ibp->

.end(); ++gfp) {

3415 if

((gfp->from >= from && gfp->from <= to) ||

3416

(gfp->to >= from && gfp->to <= to) ||

3417

(gfp->from <= from && gfp->to >= to)) {

3419 "Gap features at \"{}..{}\" and \"{}..{}\" overlap."

, from, to, gfp->from, gfp->to);

3421

}

else if

(to + 1 == gfp->from || from - 1 == gfp->to) {

3430 "Gap features at \"{}..{}\" and \"{}..{}\" are contiguous, and should probably be represented by a single gap that spans both."

, from, to, gfp->from, gfp->to);

3436 auto

tgfp = ibp->

.before_begin();

3438 auto const

nxt =

(tgfp);

3439 if

(nxt == ibp->

.end() || nxt->from >= from) {

3444

gfp = ibp->

.emplace_after(tgfp);

3448

gfp->estimated_length = estimated_length;

3450

gfp->assembly_gap =

true

;

3452

gfp->gap_type = gap_type;

3453

gfp->asn_gap_type = asn_gap_type;

3455 if

(! asn_linkage_evidence.empty()) {

3456

gfp->asn_linkage_evidence.swap(asn_linkage_evidence);

3457

asn_linkage_evidence.clear();

3461

linkage_evidence_names.clear();

3462

asn_linkage_evidence.clear();

3466 if

(ibp->

.empty())

3470

ibp->

.clear();

3477 if

(! entry || xil.empty())

3480 for

(

const auto

& xip : xil) {

3481 if

(xip.subtags.empty())

3485 for

(

const auto

& xipqual : xip.subtags) {

3497

quals.push_back(qual);

3506 if

(! entry || xil.empty())

3509 auto

xip = xil.cbegin();

3510 for

(; xip != xil.cend(); ++xip)

3514 if

(xip == xil.cend() || xip->subtags.empty())

3518 auto

dbp = dbl.before_begin();

3520 const auto

& subtags = xip->subtags;

3521 for

(

const auto

& xip : subtags) {

3522 if

(xip.subtags.empty())

3526 for

(

const auto

& xipfeat : xip.subtags) {

3534

dbp = dbl.emplace_after(dbp, 0);

3535

dbp->SetFeatData(fbp);

3539

p.mData = std::move(dbl);

3557 for

(

auto

& cur : quals) {

3558 if

(! cur->IsSetQual() || ! cur->IsSetVal())

3561 const string

& cur_qual = cur->GetQual();

3562 const string

& cur_val = cur->GetVal();

3564 if

(cur_qual !=

"note"

|| cur_val.empty())

3569 buf

.reserve(cur_val.size());

3571 for

(

const char

* cp = cur_val.c_str(); *cp !=

'\0'

; ++cp) {

3572 buf

.push_back(*cp);

3573 if

(*cp ==

';'

&& (cp[1] ==

' '

|| cp[1] ==

';'

)) {

3574 for

(++cp; *cp ==

' '

|| *cp ==

';'

;)

3577 buf

.push_back(

' '

);

3584 size

+= cur->GetVal().size();

3585 for

(

char

c : cur->GetVal())

3594

note.reserve(

size

- 1);

3596 for

(TQualVector::iterator cur = quals.begin(); cur != quals.end();) {

3597 if

(! (*cur)->IsSetQual() || ! (*cur)->IsSetVal()) {

3602 const string

& cur_qual = (*cur)->GetQual();

3603 const string

& cur_val = (*cur)->GetVal();

3605 if

(cur_qual !=

"note"

) {

3610 if

(! cur_val.empty()) {

3613 if

(! note.empty()) {

3614

note.push_back(

';'

);

3615

note.push_back(

'~'

);

3618 for

(

char

c : cur_val) {

3620

note.push_back(

'~'

);

3625

cur = quals.erase(cur);

3629

qual_new->

(

"note"

);

3630

qual_new->

SetVal

(note);

3632

quals.push_back(qual_new);

3639 for

(

char

c :

) {

3642 if

(

isalpha

'_'

)

3661 for

(

char

& c :

)

3662 if

(c >=

'A'

&& c <=

'Z'

)

3669 if

(val_str.size() < 2)

3674 for

(

size_t i

= 1;

< val_str.size(); ++

) {

3675 if

(val_str[

- 1] ==

','

&& val_str[

] !=

' '

)

3679

val_str.reserve(val_str.size() + v.size());

3681 while

(! v.empty()) {

3682 size_t i

= v.back();

3685

val_str.insert(

, 1,

' '

);

3691 const string

3692

vector<string>& lines)

3716 string

bstr(bptr, eptr);

3719

vector<string> qualLines;

3722 string

qualKey, qualVal;

3723 string

featKey(fbp->

);

3724 string

featLocation(*fbp->

);

3726 while

(! qualParser.

Done

()) {

3732

pQual->

SetVal

(qualVal);

3733

fbp->

.push_back(pQual);

3745 auto n

.find(

':'

);

3749 "/satellite qualifier \"{}\" does not begin with a valid satellite type."

);

3751

}

else if

(

!= string_view::npos &&

+ 1 >=

.size()) {

3753 "/satellite qualifier \"{}\" does not include a class or identifier after the satellite type."

);

3764 bool

found =

false

;

3767 for

(TQualVector::iterator qual = fbp->

.begin(); qual != fbp->

.end(); ++qual)

3769 if

((*qual)->IsSetQual() && (*qual)->GetQual() ==

"mobile_element_type"

3770

(*qual)->IsSetVal() && !(*qual)->GetVal().empty()) {

3771

p_val = (

char

*) (*qual)->GetVal().c_str();

3772 for

(p = p_val; *p ==

'\"'

;)

3784

optional<string> loc_str = fbp->

;

3787 "Mandatory qualifier /mobile_element_type is absent or has no value : Feature \"mobile_element\" : Location \"{}\". Feature dropped."

, loc_str.has_value() ? loc_str.value() :

"unknown"

);

3790 "Mandatory qualifier /mobile_element_type is absent or has no value : Feature \"mobile_element\" : Location \"{}\". Entry dropped."

, loc_str.has_value() ? loc_str.value() :

"unknown"

);

3803

optional<string> loc_str = fbp->

;

3806 "The value \"{}\" of qualifier /mobile_element_type is invalid for the feature \"mobile_element\" at \"{}\". Feature dropped."

, p_val, loc_str.has_value() ? loc_str.value() :

"unknown"

);

3809 "The value \"{}\" of qualifier /mobile_element_type is invalid for the feature \"mobile_element\" at \"{}\". Entry dropped."

, p_val, loc_str.has_value() ? loc_str.value() :

"unknown"

);

3856

loc =

"1.."

+ to_string(ibp->

bases

);

3858 for

(

auto

& dbp : dbl) {

3862

dbp.SetFeatData(fbp);

3864

bptr = dbp.mBuf.ptr;

3865

eptr = bptr + dbp.mBuf.len;

3867 for

(p = bptr; *p !=

'\n'

;)

3871 if

(*ptr1 ==

' '

) {

3874 for

(ptr1 = bptr; *ptr1 ==

' '

;)

3877 for

(ptr2 = ptr1; *ptr2 !=

' '

&& *ptr2 !=

'\n'

;)

3882

fbp->

"misc_feature"

;

3886 for

(ptr1 = ptr2; *ptr1 ==

' '

;)

3888 if

(*ptr1 ==

'\n'

) {

3889 if

(ibp->

is_mga

false

) {

3903 for

(ptr2 = ptr1; *ptr2 !=

'/'

&& ptr2 < eptr;)

3905 string

tmp_loc(ptr1, ptr2);

3909 for

(

char

c : tmp_loc)

3910 if

(c !=

' '

&& c !=

'\n'

)

3913

fbp->

= std::move(tmp_loc);

3920 if

(fbp->

"allele"

|| fbp->

"mutation"

) {

3922 "Obsolete feature \"{}\" found. Replaced with \"variation\"."

, fbp->

);

3923

fbp->

"variation"

;

3939 if

(fbp->

"assembly_gap"

) {

3940 for

(

const auto

& cur : fbp->

) {

3941 const string

& cur_qual = cur->GetQual();

3942 if

(cur_qual ==

"gap_type"

3943

cur_qual ==

"assembly_evidence"

) {

3945 "Qualifier /{} is invalid for the feature \"{}\" at \"{}\"."

, cur_qual, fbp->

(fbp,

"Unknown"

));

3951 if

(fbp->

"source"

) {

3952 for

(

const auto

& cur : fbp->

) {

3953 const string

& cur_qual = cur->GetQual();

3954 if

(cur_qual ==

"submitter_seqid"

) {

3956 "Qualifier /{} is invalid for the feature \"{}\" at \"{}\"."

, cur_qual, fbp->

(fbp,

"Unknown"

));

3965 if

(fbp->

"mobile_element"

3990 if

(fbp->

"mobile_element"

) {

3993 if

((fbp->

"old_sequence"

&& fbp->

"conflict"

) ||

3994

(

"citation"

)) {

3996 "lacks required /{} qualifier : feature has been dropped."

);

4003

}

else if

(fbp->

"misc_feature"

&& fbp->

.empty()) {

4012 for

(

auto

& cur : fbp->

) {

4013 if

(! cur->IsSetQual() || ! cur->IsSetVal())

4016 const string

& qual_str = cur->GetQual();

4017 string

val_str = cur->GetVal();

4021 if

(val_str.empty() && qual_str !=

"replace"

) {

4024 if

(qual_str ==

"replace"

)

4026

cur->SetVal(val_str);

4029 if

(qual_str ==

"satellite"

)

4043 if

(! fbp || fbp->

.empty())

4046 for

(TQualVector::iterator cur = fbp->

.begin(); cur != fbp->

.end();) {

4047 const string

& qual_str = (*cur)->GetQual();

4049 if

((*cur)->IsSetVal()) {

4050 string

val_str = (*cur)->GetVal();

4052 if

(qual_str ==

"translation"

) {

4054

}

else if

(qual_str ==

"rpt_unit"

) {

4056

}

else if

(qual_str ==

"cons_splice"

) {

4058

}

else if

(qual_str ==

"note"

) {

4060 for

(

size_t

p = 0;;) {

4061

p = val_str.find(

'/'

, p);

4062 if

(p == string::npos)

4065 if

CheckLegalQual

(string_view(val_str).substr(p),

' '

nullptr

))

4068 string

_loc(val_str.substr(0, 34));

4069 if

(_loc.size() > 30) {

4075 "/note qualifier value appears to contain other qualifiers : qualifier has been dropped : [{}]."

, _loc);

4079 "/note qualifier value appears to contain other qualifiers : [{}]."

, _loc);

4083

cur = fbp->

.erase(cur);

4089 for

(

char

c : val_str) {

4090 if

(c ==

'\"'

|| c ==

' '

|| c ==

'\t'

)

4097 if

(qual_str ==

"replace"

)

4098

(*cur)->SetVal(

""

);

4102

(*cur)->SetVal(val_str);

4106 if

(qual_str == *

)

4110 if

(! (*cur)->IsSetVal()) {

4111 if

(qual_str ==

"old_locus_tag"

) {

4113 "Feature \"{}\" at \"{}\" has an /old_locus_tag qualifier with no value. Qualifier has been dropped."

(fbp,

"Unknown"

(fbp,

"Empty"

));

4116 "Qualifier /{} ignored because it lacks a data value. Feature \"{}\", location \"{}\"."

, qual_str,

(fbp,

"Unknown"

(fbp,

"Empty"

));

4118

cur = fbp->

.erase(cur);

4121

}

else if

((*cur)->IsSetVal()) {

4123 "Qualifier /{} should not have data value. Qualifier value has been ignored. Feature \"{}\", location \"{}\"."

, qual_str,

(fbp,

"Unknown"

(fbp,

"Empty"

));

4127 if

((*cur)->IsSetVal() && qual_str ==

"note"

) {

4128 string val

= (*cur)->GetVal();

4129

std::replace(

.begin(),

.end(),

'\"'

'\''

);

4130

(*cur)->SetVal(

);

4147 for

(

auto

& dbp : dbl) {

4148 if

(! dbp.hasData())

4150

fbp = dbp.GetFeatData();

4154 if

(fbp->

"-"

) {

4156

fbp->

"misc_feature"

;

4159 if

(fbp->

"allele"

|| fbp->

"mutation"

) {

4161 "Obsolete feature \"{}\" found. Replaced with \"variation\"."

, fbp->

);

4162

fbp->

"variation"

;

4181 if

(fbp->

"NON_STD"

)

4182

fbp->

"MOD_RES"

;

4185 if

(keyindx < 0 && ! deb) {

4194 if

(! fbp->

.empty()) {

4199 if

(fbp->

"mobile_element"

4213

}

else if

(fbp->

spindex

< 0) {

4225 if

(fbp->

"mobile_element"

) {

4228 if

((fbp->

"old_sequence"

&& fbp->

"conflict"

) ||

4229

(

"citation"

)) {

4231 "lacks required /{} qualifier : feature has been dropped."

);

4238

}

else if

(fbp->

"misc_feature"

&& fbp->

.empty()) {

4247 for

(

auto

& cur : fbp->

) {

4248 if

(! cur->IsSetQual() || ! cur->IsSetVal())

4251 const string

& qual_str = cur->GetQual();

4252 string

val_str = cur->GetVal();

4256 if

(val_str.empty() && qual_str !=

"replace"

) {

4259 if

(qual_str ==

"replace"

)

4261

cur->SetVal(val_str);

4275 for

(

const auto

& qual : feat.

GetQual

()) {

4276 if

(! qual->IsSetQual() || qual->GetQual().empty() ||

4277

qual->GetQual() !=

"ncRNA_class"

)

4282 if

(! qual->IsSetVal() || qual->GetVal().empty()) {

4285 "Feature \"ncRNA\" at location \"{}\" has an empty /ncRNA_class qualifier."

, loc.empty() ?

"unknown"

s : loc);

4293 "Feature \"ncRNA\" at location \"{}\" has an invalid /ncRNA_class qualifier: \"{}\"."

, loc.empty() ?

"unknown"

s : loc, qual->GetVal());

4307 "Feature \"ncRNA\" at location \"{}\" {} /ncRNA_class qualifier."

, loc.empty() ?

"unknown"

s : loc, (

count

== 0) ?

"lacks the mandatory"

"has more than one"

);

4315 for

(

auto

qual = feat.

().begin(); qual != feat.

().end(); ++qual) {

4316 if

(! (*qual)->IsSetQual() || (*qual)->GetQual() !=

"artificial_location"

)

4319 if

((*qual)->IsSetVal()) {

4320 const Char

* p_val = (*qual)->GetVal().c_str();

4321 for

(; *p_val ==

'\"'

;)

4324 if

(*p_val ==

'\0'

)

4325

(*qual)->ResetVal();

4328 string val

= (*qual)->IsSetVal() ? (*qual)->GetVal() :

""

;

4330 if

(

"heterogenous population sequenced"

4331 val

"low-quality sequence region"

) {

4338

except_text +=

", "

;

4339

except_text +=

;

4345 msg

(

"Encountered empty /artificial_location qualifier : Feature \"{}\" : Location \"{}\". Qualifier dropped."

.empty() ?

"unknown"

s :

, loc_str.empty() ?

"unknown"

s : loc_str);

4347 msg

(

"Value \"{}\" is not legal for the /artificial_location qualifier : Feature \"{}\" : Location \"{}\". Qualifier dropped."

.empty() ?

"unknown"

s :

, loc_str.empty() ?

"unknown"

s : loc_str);

4377 if

(fbp1->

.empty() && ! fbp2->

.empty())

4379 if

(! fbp1->

.empty() && fbp2->

.empty())

4381 if

(! fbp1->

.empty() && ! fbp2->

.empty()) {

4399 return

fbp1->

< fbp2->

;

4405 if

(! fbp || fbp->

.empty() || ! rclass)

4408

fbp->

"regulatory"

;

4411

qual->

(

"regulatory_class"

);

4413

fbp->

.push_back(qual);

4425 for

(

auto

& dbp : dbl) {

4426

fbp = dbp.GetFeatData();

4427 if

(! fbp || fbp->

.empty())

4430 if

(fbp->

"attenuator"

)

4432 else if

(fbp->

"CAAT_signal"

)

4434 else if

(fbp->

"enhancer"

)

4436 else if

(fbp->

"GC_signal"

)

4438 else if

(fbp->

"-35_signal"

)

4440 else if

(fbp->

"-10_signal"

)

4442 else if

(fbp->

"polyA_signal"

)

4444 else if

(fbp->

"promoter"

)

4446 else if

(fbp->

"RBS"

)

4448 else if

(fbp->

"TATA_signal"

)

4450 else if

(fbp->

"terminator"

)

4452 else if

(fbp->

"regulatory"

)

4456

other_class =

false

;

4459 for

(

const auto

& cur : fbp->

) {

4460 if

(! cur->IsSetQual() || ! cur->IsSetVal())

4463 const string

& qual_str = cur->GetQual();

4465 if

(qual_str !=

"regulatory_class"

) {

4466 if

(qual_str ==

"note"

)

4472 if

(! cur->IsSetVal() || cur->GetVal().empty()) {

4479 "Empty /regulatory_class qualifier value in regulatory feature at location {}."

, s);

4484 const string

& val_str = cur->GetVal();

4487 if

(val_str == *

)

4491 if

(val_str ==

"other"

)

4492

other_class =

true

;

4502 "Invalid /regulatory_class qualifier value {} provided in regulatory feature at location {}."

, val_str, s);

4513 "The regulatory feature is missing mandatory /regulatory_class qualifier at location {}."

, s);

4515

}

else if

(

count

> 1) {

4525 if

(other_class && ! got_note) {

4539 const string

& submitter_seqid,

4544 if

(seqtype == 0 || seqtype == 1 || seqtype == 7)

4546 else if

(seqtype == 4 || seqtype == 5 || seqtype == 8 || seqtype == 9)

4554 tag

.SetTag().SetStr(submitter_seqid);

4556

bioseq.

SetId

().push_back(gen_id);

4575 if

(seqtype == 0 || seqtype == 3 || seqtype == 4 || seqtype == 6 ||

4576

seqtype == 10 || seqtype == 12) {

4582 if

(seqtype == 1 || seqtype == 5 || seqtype == 7 || seqtype == 8 ||

4583

seqtype == 9 || seqtype == 11) {

4585 if

(prefix[4] >=

'0'

&& prefix[4] <=

'9'

)

4595 for

(

auto

tbp = ibp->

secaccs

.begin(); tbp != ibp->

secaccs

.end(); ++tbp) {

4596 if

((*tbp)[0] ==

'-'

)

4602 i

= (prefix[4] >=

'0'

&& prefix[4] <=

'9'

) ? 6 : 8;

4610 if

(

&& prefix) {

4612 if

(seqtype == 0 || seqtype == 1 || seqtype == 4 || seqtype == 5 ||

4613

seqtype == 7 || seqtype == 8 || seqtype == 9 || seqtype == 10 ||

4615 if

(prefix[4] >=

'0'

&& prefix[4] <=

'9'

)

4632

CDelta_ext::Tdata::iterator

delta

;

4635 const CSeq_id

id

nullptr

;

4637 if

(! (*delta)->IsLoc())

4640 const CSeq_loc

& locs = (*delta)->GetLoc();

4645 if

(! loc->

IsInt

())

4655 const CTextseq_id

* text_id =

id

->GetTextseq_Id();

4664 i

= (prefix[4] >=

'0'

&& prefix[4] <=

'9'

) ? 6 : 8;

4673 if

(

delta

== deltas.end() && prefix) {

4675 if

(seqtype == 0 || seqtype == 1 || seqtype == 4 || seqtype == 5 ||

4676

seqtype == 7 || seqtype == 8 || seqtype == 9 || seqtype == 10 ||

4678 if

(prefix[4] >=

'0'

&& prefix[4] <=

'9'

)

4699 "Submitter sequence identifiers for non-project-based TSA records are not supported. /submitter_seqid \"{}\" has been dropped."

, ibp->

submitter_seqid

);

4704 "Only WGS/TLS/TSA related records (contigs and scaffolds) are allowed to have /submitter_seqid qualifier. This \"{}\" is not one of them. Entry dropped."

, ibp->

acnum

);

4713 auto

& descrList = descrs.

Set

();

4714 auto

it = descrList.begin();

4715 while

(it != descrList.end()) {

4716 if

((*it)->IsSource()) {

4717

it = descrList.erase(it);

4784

dab = temp_xml_chain.begin();

4785

dab_end = temp_xml_chain.end();

4788

dab = chain.begin();

4789

dab_end = chain.end();

4791 while

(dab != dab_end && dab->mType !=

)

4795 for

(

auto

dbp = dab; dbp != dab_end; ++dbp) {

4797 if

(dblk.mType !=

)

4803 TDataBlkList

& dbl = std::get<TDataBlkList>(dblk.mData);

4814 for

(

auto

& tdbp : dbl) {

4819 for

(

auto

& tdbp : dbl)

4828 if

(

> 1 && ibp->

is_mga

) {

4837 if

(! ibp->

drop

) {

4843

p.erase(

remove

(p.begin(), p.end(),

' '

), p.end());

4844

p.erase(

remove

(p.begin(), p.end(),

'\t'

), p.end());

4850 if

(seq_feats.empty()) {

4852 for

(; dab != dab_end; ++dab) {

4853 if

(dab->hasData()) {

4854 TDataBlkList

& dbl = std::get<TDataBlkList>(dab->mData);

4856

dab->mData = monostate();

4860

temp_xml_chain.clear();

4875

descr_src->

SetSource

(seq_feats.front()->SetData().SetBiosrc());

4876

bioseq.

SetDescr

().Set().push_back(descr_src);

4880 const CBioSource

& bio = seq_feats.front()->GetData().GetBiosrc();

4883 if

(! taxname.empty())

4885 "BioSource descriptor and Source dropped because \"{}\" is not present in the NCBI taxonomy database."

, taxname);

4888 "BioSource descriptor and Source dropped because provided organism is not present in the NCBI taxonomy database."

);

4891

seq_feats.pop_front();

4893 for

(; dab != dab_end; ++dab) {

4894 if

(dab->mType !=

) {

4898 TDataBlkList

& dbl = std::get<TDataBlkList>(dab->mData);

4899 for

(

auto

& dbp : dbl) {

4900 if

(dbp.mDrop ==

true

)

4904 if

(fbp->

"source"

4905

fbp->

"assembly_gap"

4906

(fbp->

"gap"

4916 if

(feat.

()) {

4917 if

(fbp->

"CDS"

) {

4919 "CDS feature has unparsable location. Entry dropped. Location = [{}]."

, *fbp->

);

4930 "Location pointing outside the entry [{}]"

, *fbp->

);

4934 if

(imp_feat.

() ==

"intron"

4935

imp_feat.

() ==

"exon"

) {

4951

seq_feats.push_back(feat);

4960 "Mixed strands in SeqLoc of /trans_splicing feature: {}"

, *fbp->

);

4963 "Mixed strands in SeqLoc: {}"

, *fbp->

);

4967

seq_feats.push_back(feat);

4973

temp_xml_chain.clear();

4983 for

(

auto

& feat : seq_feats) {

4984 if

(! feat->GetData().IsImp())

4987 const CImp_feat

& imp_feat = feat->GetData().GetImp();

4990

imp_feat.

().find(

"RNA"

) != string::npos) {

5007 SeqFeatPub

(pp, entry, seq_feats, *seq_id, col_data, ibp);

5008 if

(seq_feats.empty() && ibp->

drop

) {

5015 ImpFeatPub

(pp, entry, seq_feats, *seq_id, col_data, ibp);

5018 if

(seq_feats.empty())

5022

annot->

().SetFtable().swap(seq_feats);

5024

bioseq.

SetAnnot

().push_back(annot);

5030 char

* p =

nullptr

;

5034 if

(! p || (tRNA && tRNA < p))

5036 if

(! p || (rRNA && rRNA < p))

5038 if

(! p || (snRNA && snRNA < p))

5040 if

(! p || (scRNA && scRNA < p))

5042 if

(! p || (uRNA && uRNA < p))

5044 if

(! p || (snoRNA && snoRNA < p))

5053 if

(p == snRNA || p == uRNA)

5075 char

* mRNA =

nullptr

;

5076 char

* tRNA =

nullptr

;

5077 char

* rRNA =

nullptr

;

5078 char

* snRNA =

nullptr

;

5079 char

* scRNA =

nullptr

;

5080 char

* uRNA =

nullptr

;

5081 char

* snoRNA =

nullptr

;

5107 if

(! ibp->

.empty()) {

5118 while

';'

&& *

'\n'

&& *

'\0'

)

5121 while

';'

&& *

' '

&& *

'\t'

&& *

'\n'

5124 if

(

- molstr > 10)

5130

q = (

char

"???"

;

5132

q = (

char

"???"

;

5135 if

(ibp->

"genomic DNA"

) {

5145

}

else if

(ibp->

"genomic RNA"

) {

5154

}

else if

(ibp->

"mRNA"

) {

5163

}

else if

(ibp->

"tRNA"

) {

5172

}

else if

(ibp->

"rRNA"

) {

5181

}

else if

(ibp->

"snoRNA"

) {

5190

}

else if

(ibp->

"snRNA"

) {

5199

}

else if

(ibp->

"scRNA"

) {

5208

}

else if

(ibp->

"pre-RNA"

) {

5217

}

else if

(ibp->

"pre-mRNA"

) {

5226

}

else if

(ibp->

"other RNA"

) {

5238

}

else if

(ibp->

"other DNA"

) {

5250

}

else if

(ibp->

"unassigned RNA"

) {

5262

}

else if

(ibp->

"unassigned DNA"

) {

5274

}

else if

(ibp->

"viral cRNA"

) {

5285

}

else if

(ibp->

"transcribed RNA"

) {

5300 "Invalid /mol_type value \"{}\" provided in source features. Entry dropped."

, ibp->

);

5310 "Molecule type \"{}\" from the ID line disagrees with \"{}\" from the /mol_type qualifier."

, q, ibp->

);

5317 "Molecule type \"{}\" from the ID/LOCUS line disagrees with \"{}\" from the /mol_type qualifier."

, q, ibp->

);

5324

ibp->

"genomic DNA"

)

5353 "Molecule type \"{}\" from the /mol_type qualifier disagrees with this record's sequence type: \"{}\"."

, ibp->