#
#  converstion script to transform a darwin Genome DB into
#  an SeqXML conformant file.
#
#  SeqXML emerged from the Quest for Orthologs conference in 2009.
#
#  schema file is available at http://seqxml.org/0.2/seqxml.xsd
#  
#                                      Adrian A. August 2010

WriteSeqXML := proc(fn:string)
    global DB;
    if nargs<1 then 
        error('no output filename assigned'); 
    elif nargs>1 then
        if type(DB,database) then curDB := DB: fi:
        if type(args[2],string) then
            DB := ReadDb(args[2]);
        elif type(args[2],database) then 
            DB := args[2];
        fi:
    elif not type(DB,database) then 
        error(args,'no database specified'); 
    fi:
    seqXMLversion := '0.2';
    rel := SearchTag('DBRELEASE',DB['string']);  # default release tag
    srcTag := SearchTag('SOURCE', DB['string']);
    dbnTag := SearchTag('DBNAME', DB['string']);
    cmnTag := SearchTag('COMMENT',DB['string']);
    code := SearchTag('5LETTERNAME', DB['string']);
    sci := SearchTag('SCINAME',DB['string']);
    tax := SearchTag('TAXONID',DB['string']);
    if tax='' then 
       tax := traperror(TaxonId(code));
       if tax=lasterror then tax := '0';
       else tax := string(tax) fi:
    fi:

    if SearchString('genome reviews',cmnTag)>0 then
        src := 'Genome Reviews';
    elif SearchString('ebi.ac.uk',srcTag)>0
          or cmnTag[1..6]='\nID   ' then
        src := 'EBI';
    elif SearchString('ensembl.org', srcTag)>0 then
        src := 'Ensembl';
    elif SearchString('ensemblgenomes.org', srcTag)>0 then
        src := 'EnsemblGenomes';
    elif SearchString('ncbi.nlm.nih.gov', srcTag)>0
           and cmnTag[1..6]='\nLOCUS' then
        src := 'NCBI';
    elif SearchString('from NCBI', dbnTag)>0 then
        src := 'NCBI';
    elif SearchString('from JGI', dbnTag)>0 then
        src := 'JGI';
    elif SearchString('from Phytozome', dbnTag)>0 then
        src := 'Phytozome';
    elif SearchString('BROAD Institute', dbnTag)>0 then
        src := 'BROAD Institute';
    elif SearchString('from Wormbase', dbnTag)>0 then
        src := 'WormBase';
    elif SearchString('flybase.net', srcTag)>0 then
        src := 'FlyBase';
    elif SearchString('dictyBase', dbnTag)>0 then
        src := 'dictyBase';
    elif SearchString('from GiardiaDB', dbnTag)>0 then
        src := 'GiardiaDB';
    elif SearchString('from the International Rice Genome Sequencing Project', dbnTag)>0 then
        src := 'Int. Rice Genome Sequencing Project';
    elif SearchString(' from ', dbnTag)>0 then
        src := dbnTag[ SearchString('from',dbnTag)+6 .. -1];
    else src := 'unknown'; lprint('WARNING: unknown genome source'); fi:

    refFields := ['Uniprot/SWISSPROT','UniprotKB/Swiss-Prot', 
                  'Uniprot/SPTREMBL', 'IPI', 'RefSeq_peptide'];
    verb := Set(printgc=false);
    OpenWriting(fn):
    printf('<?xml version="1.0"?>\n');
    printf('<seqXML source="%s" sourceVersion="%s" seqXMLversion="%s">\n', 
            src, rel, seqXMLversion);
    for i to DB[TotEntries] do
        e := Entry(i);
        printf(' <entry id="%s%05d">\n',code,i);
        printf('  <species name="%s" ncbiTaxID="%s"/>\n', sci, tax);
        printf('  <AAseq>%s</AAseq>\n', Sequence(e));
        for tag in refFields do
            id := SearchTag(tag, e):
            if id='' then next fi;
            printf('  <DBRef type="%s" source="%s" id="%s"/>\n', 'AA', tag, id);
        od:
        id := SearchDelim('; ', SearchTag('ID',e))[1]:
        ac := SearchDelim('; ', SearchTag('AC',e)):
        for z in {id, op(ac)} do
            printf('  <DBRef type="%s" source="%s" id="%s"/>\n', 'AA', src, z);
        od:
        printf(' </entry>\n');
    od:
    printf('</seqXML>\n');
    OpenWriting(previous):
    Set(printgc=verb);
    if type(curDB,database) then DB := curDB fi:
end: