RetroSearch Browse

;

304

m_ObjMgr(0), m_Continue(

false

305

m_Level(0), m_ReportStream(0), m_NeedReportHeader(

true

), m_AsnOut(0),

306

m_LogStream(0), m_Mode(e_report_diffs), m_ReturnCode(0),

307

m_StructuredCommentPrefix(

""

), m_CompareStructuredComments(

true

308

m_FirstSeqOnly(

false

), m_IDPrefix(

""

), m_HUPDate(

""

309

m_BioSampleAccession(

""

), m_BioProjectAccession(

""

310

m_Owner(

""

), m_Comment(

""

311

m_Processed(0), m_Unprocessed(0), m_Handler(

NULL

)

324

arg_desc->AddOptionalKey

325

(

"p"

"Directory"

"Path to ASN.1 Files"

327

arg_desc->AddOptionalKey

328

(

"i"

"InFile"

"Single Input File"

330

arg_desc->AddOptionalKey(

331 "o"

"OutFile"

"Single Output File"

333

arg_desc->AddOptionalKey(

334 "f"

"Filter"

"Substring Filter"

336

arg_desc->AddDefaultKey

338

arg_desc->AddFlag(

"u"

"Recurse"

);

339

arg_desc->AddFlag(

"d"

"Use development Biosample server"

);

341

arg_desc->AddDefaultKey(

"a"

342 "ASN.1 Type (a Automatic, z Any, e Seq-entry, b Bioseq, s Bioseq-set, m Seq-submit, t Batch Bioseq-set, u Batch Seq-submit) or accession list (l)"

346

arg_desc->AddFlag(

"b"

"Output binary ASN.1"

);

348

arg_desc->AddFlag(

"M"

"Process only first sequence in file (master)"

);

352

arg_desc->AddOptionalKey(

353 "L"

"OutFile"

"Log File"

356

arg_desc->AddDefaultKey(

357 "m"

"mode"

"Mode:\n" 358 "\t1 create update file\n" 359 "\t2 generate file for creating new biosample entries\n" 360 "\t3 push source info from one file (-i) to others (-p)\n" 361 "\t4 update with source qualifiers from BioSample unless conflict\n" 362 "\t5 update with source qualifiers from BioSample (continue with conflict))\n" 363 "\t6 report transaction status\n" 364 "\t7 use web API for update (with delete)\n" 365 "\t8 use web API for update (no delete)\n"

368

arg_desc->SetConstraint(

"m"

, constraint);

370

arg_desc->AddOptionalKey(

373

arg_desc->AddOptionalKey(

374 "biosample"

"BioSampleAccession"

"BioSample Accession to use for sequences in record. Report error if sequences contain a reference to a different BioSample accession."

);

375

arg_desc->AddOptionalKey(

376 "bioproject"

"BioProjectAccession"

"BioProject Accession to use for sequences in record. Report error if sequences contain a reference to a different BioProject accession."

);

377

arg_desc->AddOptionalKey(

"comment"

"BioSampleComment"

"Comment to use for creating new BioSample xml"

);

379

arg_desc->AddOptionalKey(

"apikey_file"

"BioSampleWebAPIKey"

"File containing Web API Key needed to update BioSample database"

);

382 string

prog_description =

"BioSample Checker\n"

;

383

arg_desc->SetUsageContext(

GetArguments

().GetProgramBasename(),

384

prog_description,

false

);

399 string

header =

->ReadFileHeader();

401 bool

unhandled =

false

;

403 if

(header ==

"Seq-submit"

) {

405

}

else if

( header ==

"Seq-entry"

) {

407

}

else if

(header ==

"Bioseq-set"

) {

440 while

( !

->AtEOF() ) {

450

"Unable to fetch Bioseq for "

<< line << endl;

452 id

->GetLabel(&

label

);

478 while

( !

->AtEOF() ) {

492 bool

need_to_close_report =

false

;

493 bool

need_to_close_asn =

false

;

500 if

(pos != string::npos) {

501

path = path.substr(0, pos);

503

path = path +

".val"

;

511

need_to_close_report =

true

;

520 if

(pos != string::npos) {

521

path = path.substr(0, pos);

523

path = path +

".out"

;

525

need_to_close_asn =

true

;

538 if

(

.get() ==

nullptr

) {

541 if

->InGoodState()) {

563 if

(need_to_close_report) {

573 if

(need_to_close_asn) {

592

vector<CRef<CSeqdesc> > descriptors;

601

src_desc->

Assign

(*src_desc_ci);

602

descriptors.push_back(src_desc);

612

vector<CRef<CSeqdesc> > descriptors;

637 string

header =

->ReadFileHeader();

639

vector<CRef<CSeqdesc> > descriptors;

640 if

(header ==

"Seq-submit"

) {

642

}

else if

( header ==

"Seq-entry"

) {

654 int

num_of_files = 0;

658 for

(

const auto

&ii : files) {

659 string

fname = ii->GetName();

661

(!file_suffix.empty() ||

NStr::Find

(fname, file_suffix) != string::npos)) {

669 for

(

const auto

&ii : subdirs) {

670 string

subdir = ii->GetName();

690 m_Mode

= args[

"m"

].AsInteger();

692 m_IDPrefix

= args[

"R"

] ? args[

"R"

].AsString() :

""

;

693 m_HUPDate

= args[

"HUP"

] ? args[

"HUP"

].AsString() :

""

;

696 m_Comment

= args[

"comment"

] ? args[

"comment"

].AsString() :

""

;

698 string

apikey_file = args[

"apikey_file"

] ? args[

"apikey_file"

].AsString() :

""

;

699 if

(!apikey_file.empty()) {

700

ifstream is(apikey_file.c_str());

725 SaveFile

(args[

"o"

].AsString(), args[

"b"

]);

752

"Structured comment prefix is only appropriate for generating a biosample table."

<< endl;

767

}

else if

(

NStr::Equal

(args[

"a"

].AsString(),

"f"

)) {

773 string

dir_name = (args[

"p"

]) ? args[

"p"

].AsString() :

""

;

774 string

file_suffix = (args[

"f"

]) ? args[

"f"

].AsString() :

""

;

775 string

file_mask = (args[

"x"

]) ? args[

"x"

].AsString() :

".sqn"

;

776

file_mask =

"*"

+ file_mask;

777 bool

dir_recurse = args[

"u"

];

788

"List type (-a l or -a f) is not appropriate for push mode."

<< endl;

790

}

else if

(!args[

"p"

] || !args[

"i"

]) {

792

"Both directory containing contigs (-p) and master file (-i) are required for push mode."

<< endl;

798

}

else if

( args[

"p"

] ) {

864

}

catch

(std::exception e) {

872 in

.ReadClassMember(member);

880

(

const CArgs

& args)

914 if

(

->GetNum_rows() == 0) {

919 for

(

const auto

&it :

->GetColumns()) {

923 for

(

size_t row

= 0;

row

< (size_t)

->GetNum_rows();

row

++) {

924 for

(

const auto

&it :

->GetColumns()) {

925 if

(row < it->GetData().GetString().size()) {

946

biosample_util::CBiosampleFieldDiff::PrintHeader(*

m_ReportStream

false

);

950 for

(

const auto

&it : diffs) {

972

vector< CRef<biosample_util::CBiosampleFieldDiff> >

add_item

;

973

vector< CRef<biosample_util::CBiosampleFieldDiff> > change_item;

974

vector< CRef<biosample_util::CBiosampleFieldDiff> > delete_item;

975

vector< CRef<biosample_util::CBiosampleFieldDiff> > change_organism;

979 for

(

const auto

&it : diffs) {

980 string id

= it->GetBioSample();

981 string smp

= it->GetSampleVal();

982 string

src = it->GetSrcVal();

983 string

fld = it->GetFieldName();

986 if

(blank_smp && blank_src) {

993 if

(fld ==

"Organism Name"

) {

994

change_organism.push_back(it);

995

}

else if

(blank_smp) {

997

}

else if

(blank_src) {

999

delete_item.push_back(it);

1002

change_item.push_back(it);

1011

options_obj.

(

"attribute_synonyms"

"true"

);

1013 for

(

auto

id

: ids) {

1015

obj1.

(

"samples"

id

);

1022

obj2.

(

"name"

, itm->GetFieldName());

1023

obj2.

(

"new_value"

, itm->GetSrcVal());

1027 if

(! delete_item.empty()) {

1030 for

(

auto

& itm : delete_item) {

1032

obj2.

(

"name"

, itm->GetFieldName());

1033

obj2.

(

"old_value"

, itm->GetSampleVal());

1037 if

(! change_item.empty() || ! change_organism.empty()) {

1039 if

(! change_organism.empty()) {

1041 for

(

auto

& itm : change_organism) {

1042

chg_org.

(

"new_value"

, itm->GetSrcVal());

1045 if

(! change_item.empty()) {

1047 for

(

auto

& itm : change_item) {

1048 string

fld = itm->GetFieldName();

1049 if

(fld ==

"Tax ID"

) {

1053

obj2.

(

"name"

, fld);

1054

obj2.

(

"old_value"

, itm->GetSampleVal());

1055

obj2.

(

"new_value"

, itm->GetSrcVal());

1061 if

( ids.

size

() > 1 ) {

1062

"ERROR: More than one BioSample ID is not supported by -m 7."

<< endl;

1071 string

sUrl =

"https://api-int.ncbi.nlm.nih.gov/biosample/update/"

;

1073

sUrl =

"https://dev-api-int.ncbi.nlm.nih.gov/biosample/update/"

;

1075 string

sContentType =

"application/json; charset=utf-8"

;

1094

vector<string> unprocessed_ids;

1104 if

(! new_diffs.empty()) {

1105 m_Diffs

.insert(

m_Diffs

.end(), new_diffs.begin(), new_diffs.end());

1106 for

(

const auto

id

: unprocessed_ids) {

1107

"Failed to retrieve BioSample data for "

id

<< endl;

1117 if

(it->IsSource()) {

1131

label

" has conflicting BioSample Accession "

<< biosample_ids[0] << endl;

1135 if

(biosample_ids.empty()) {

1140 for

(

const auto

id

: biosample_ids) {

1144 copy

(descr->

Set

().begin(), descr->

Set

().end(),

1186