/**
* Parse the header and set the values in the sequence
*
* @param header
* @param sequence
*/
public void parseHeader(String header, S sequence) {
// uniptrot
// tr|Q0TET7|Q0TET7_ECOL5 Putative uncharacterized protein OS=Escherichia coli O6:K15:H31
// (strain 536 / UPEC) GN=ECP_2553 PE=4 SV=1
sequence.setOriginalHeader(header);
String[] data = getHeaderValues(header);
if (data.length == 1) {
sequence.setAccession(new AccessionID(data[0]));
} else if (data[0].equalsIgnoreCase("sp") || data[0].equalsIgnoreCase("tr")) {
if (data[0].equalsIgnoreCase("sp")) {
sequence.setAnnotationType(AnnotationType.CURATED);
} else {
sequence.setAnnotationType(AnnotationType.PREDICTED);
}
sequence.setAccession(new AccessionID(data[1], DataSource.UNIPROT));
if (data.length > 1) {
sequence.setDescription(data[2]);
}
} else if (data[0].equalsIgnoreCase("gi")) {
DataSource giSource = DataSource.UNKNOWN;
if (data.length >= 3) {
if (data[2].equalsIgnoreCase("gb")) {
giSource = DataSource.GENBANK;
} else if (data[2].equalsIgnoreCase("emb")) {
giSource = DataSource.ENA;
} else if (data[2].equalsIgnoreCase("dbj")) {
giSource = DataSource.DDBJ;
}
sequence.setAccession(new AccessionID(data[3], giSource));
} else {
sequence.setAccession(new AccessionID(header, giSource));
}
} else if (data[0].equalsIgnoreCase("pir")) {
sequence.setAccession(new AccessionID(data[2], DataSource.NBRF));
} else if (data[0].equalsIgnoreCase("prf")) {
sequence.setAccession(new AccessionID(data[2], DataSource.PRF));
} else if (data[0].equalsIgnoreCase("pdb")) {
sequence.setAccession(new AccessionID(data[1] + ":" + data[2], DataSource.PDB1));
} else if (data[0].startsWith("PDB")) {
String[] pdbe = data[0].split(" ");
String[] pdbaccession = pdbe[0].split(":");
sequence.setAccession(new AccessionID(pdbaccession[1], DataSource.PDBe));
} else if (data[0].indexOf(":") != -1 && data.length > 1 && data[1].equals("PDBID")) {
sequence.setAccession(new AccessionID(data[0], DataSource.PDB2));
} else if (data[0].equalsIgnoreCase("pat")) {
sequence.setAccession(new AccessionID(data[2], DataSource.PATENTS));
} else if (data[0].equalsIgnoreCase("bbs")) {
sequence.setAccession(new AccessionID(data[1], DataSource.GENINFO));
} else if (data[0].equalsIgnoreCase("gnl")) {
sequence.setAccession(new AccessionID(data[2], DataSource.GENERAL));
} else if (data[0].equalsIgnoreCase("ref")) {
sequence.setAccession(new AccessionID(data[1], DataSource.NCBI));
} else if (data[0].equalsIgnoreCase("lcl")) {
sequence.setAccession(new AccessionID(data[1], DataSource.LOCAL));
} else {
sequence.setAccession(
new AccessionID(
data[0])); // avoid the common problem of picking up all the comments original header
// in getOriginalHeader
}
}
/**
* Parse the header and set the values in the sequence
*
* @param header
* @param sequence
*/
public void parseHeader(String header, S sequence) {
sequence.setOriginalHeader(header);
sequence.setAccession(new AccessionID(accession));
sequence.setDescription(description);
}