public PositionMapping(MZTabColumnFactory factory, string[] headerList)
{
for (int physicalPosition = 0; physicalPosition < headerList.Length; physicalPosition++){
string header = headerList[physicalPosition];
MZTabColumn column = factory.FindColumnByHeader(header);
if (column != null){
put(physicalPosition, column.LogicPosition);
}
}
}
/**
* Parse a header line into {@link MZTabColumnFactory} structure.
*
* @param factory SHOULD NOT set null
* @param metadata SHOULD NOT set null
*/
protected MZTabHeaderLineParser(MZTabColumnFactory factory, Metadata metadata)
{
if (factory == null){
throw new NullReferenceException("Header line should be parse first!");
}
this.factory = factory;
if (metadata == null){
throw new NullReferenceException("Metadata should be create first!");
}
this.metadata = metadata;
}
/**
* Based on factory, navigate colUnitMap<defineLabel, valueLabel>
* and refine colunit columns are correct or not.
*
* Notice: after refined phase, colunit definition can be record in the metadata.
*/
public void refineColUnit(MZTabColumnFactory factory)
{
foreach (string defineLabel in _colUnitMap.Keys){
if (defineLabel.Equals("colunit-" + Section.toDataSection(factory.Section).Name,
StringComparison.InvariantCultureIgnoreCase)){
string valueLabel = _colUnitMap[defineLabel];
MZTabError error = checkColUnit(valueLabel, factory);
if (error != null){
throw new MZTabException(error);
}
}
}
}
/**
* Generate a mzTab data line parser. A couple of common method used to parse a data line into
* {@link MZTabRecord} structure.
*
* NOTICE: {@link MZTabColumnFactory} maintain a couple of {@link MZTabColumn} which have internal logical
* position and order. In physical mzTab file, we allow user not obey this logical position organized way,
* and provide their date with own order. In order to distinguish them, we use physical position (a positive
* integer) to record the column location in mzTab file. And use {@link PositionMapping} structure the maintain
* the mapping between them.
*
* @param factory SHOULD NOT set null
* @param positionMapping SHOULD NOT set null
* @param metadata SHOULD NOT set null
*/
protected MZTabDataLineParser(MZTabColumnFactory factory, PositionMapping positionMapping,
Metadata metadata, MZTabErrorList errorList)
{
if (factory == null){
throw new NullReferenceException("Column header factory should be create first.");
}
this.factory = factory;
this.positionMapping = positionMapping;
exchangeMapping = positionMapping.exchange();
mapping = factory.GetOffsetColumnsMap();
if (metadata == null){
throw new NullReferenceException("Metadata should be parser first.");
}
this.metadata = metadata;
_errorList = errorList ?? new MZTabErrorList();
}
public SmallMolecule(MZTabColumnFactory factory, Metadata metadata)
: base(factory)
{
this.metadata = metadata;
}
public Peptide(MZTabColumnFactory factory, Metadata metadata)
: base(factory)
{
this.metadata = metadata;
}
/**
* Set the Small Molecule header line column factory.
*
* @param smallMoleculeColumnFactory if null, system will ignore Small Molecule table output.
*/
public void setSmallMoleculeColumnFactory(MZTabColumnFactory smallMoleculeColumnFactory)
{
_smallMoleculeColumnFactory = smallMoleculeColumnFactory;
}
/**
* Set the PSM header line column factory.
*
* @param psmColumnFactory if null, system will ignore PSM table output.
*/
public void setPSMColumnFactory(MZTabColumnFactory psmColumnFactory)
{
_psmColumnFactory = psmColumnFactory;
}
/**
* Set the Protein header line column factory.
*
* @param proteinColumnFactory if null, system will ignore Protein table output.
*/
public void setProteinColumnFactory(MZTabColumnFactory proteinColumnFactory)
{
_proteinColumnFactory = proteinColumnFactory;
}
/**
* Set the Peptide header line column factory.
*
* @param peptideColumnFactory if null, system will ignore Peptide table output.
*/
public void setPeptideColumnFactory(MZTabColumnFactory peptideColumnFactory)
{
_peptideColumnFactory = peptideColumnFactory;
}
public PositionMapping(MZTabColumnFactory factory, string headerLine)
: this(factory, headerLine.Split('\t'))
{
}
protected void AddOptionalColumns(MZTabColumnFactory factory, List<MZTabColumn> columns, string elementName)
{
Regex regex = new Regex(String.Format(@"(opt)_{0}\[[0-9]+\]_(.*)", elementName));
IList<string> names = columns.Select(x => x.Name).ToList();
foreach (MZTabColumn col in factory.OptionalColumnMapping.Values) {
if (regex.IsMatch(col.Name)) {
var match = regex.Match(col.Name);
string name = String.Format("{0}_{1}", match.Groups[1].Value, match.Groups[2].Value);
if (names.Contains(name)) {
continue;
}
names.Add(name);
columns.Add(new MZTabColumn(name, col.Type, col.isOptional(), col.Order));
}
}
}
private static void AddStableColumn(MZTabColumnFactory factory, MZTabColumn column)
{
factory.stableColumnMapping.Add(column.LogicPosition, column);
}
/**
* Based on {@link #section} to generate stable columns, and store them into {@link #stableColumnMapping} and
* {@link #columnMapping} which maintain all columns in the factory.
*
* @param section SHOULD be {@link Section#Protein_Header}, {@link Section#Peptide_Header} {@link Section#PSM_Header}
* or {@link Section#Small_Molecule_Header}.
*/
public static MZTabColumnFactory GetInstance(Section section)
{
section = Section.toHeaderSection(section);
if (section == null){
throw new ArgumentException(
"Section should use Protein_Header, Peptide_Header, PSM_Header or Small_Molecule_Header.");
}
MZTabColumnFactory factory = new MZTabColumnFactory();
if (section.Equals(Section.Protein_Header)){
AddStableColumn(factory, ProteinColumn.ACCESSION);
AddStableColumn(factory, ProteinColumn.DESCRIPTION);
AddStableColumn(factory, ProteinColumn.TAXID);
AddStableColumn(factory, ProteinColumn.SPECIES);
AddStableColumn(factory, ProteinColumn.DATABASE);
AddStableColumn(factory, ProteinColumn.DATABASE_VERSION);
AddStableColumn(factory, ProteinColumn.SEARCH_ENGINE);
AddStableColumn(factory, ProteinColumn.AMBIGUITY_MEMBERS);
AddStableColumn(factory, ProteinColumn.MODIFICATIONS);
AddStableColumn(factory, ProteinColumn.PROTEIN_COVERAGE);
}
else if (section.Equals(Section.Peptide_Header)){
AddStableColumn(factory, PeptideColumn.SEQUENCE);
AddStableColumn(factory, PeptideColumn.ACCESSION);
AddStableColumn(factory, PeptideColumn.UNIQUE);
AddStableColumn(factory, PeptideColumn.DATABASE);
AddStableColumn(factory, PeptideColumn.DATABASE_VERSION);
AddStableColumn(factory, PeptideColumn.SEARCH_ENGINE);
AddStableColumn(factory, PeptideColumn.MODIFICATIONS);
AddStableColumn(factory, PeptideColumn.RETENTION_TIME);
AddStableColumn(factory, PeptideColumn.RETENTION_TIME_WINDOW);
AddStableColumn(factory, PeptideColumn.CHARGE);
AddStableColumn(factory, PeptideColumn.MASS_TO_CHARGE);
AddStableColumn(factory, PeptideColumn.SPECTRA_REF);
}
else if (section.Equals(Section.PSM_Header)){
AddStableColumn(factory, PSMColumn.SEQUENCE);
AddStableColumn(factory, PSMColumn.PSM_ID);
AddStableColumn(factory, PSMColumn.ACCESSION);
AddStableColumn(factory, PSMColumn.UNIQUE);
AddStableColumn(factory, PSMColumn.DATABASE);
AddStableColumn(factory, PSMColumn.DATABASE_VERSION);
AddStableColumn(factory, PSMColumn.SEARCH_ENGINE);
AddStableColumn(factory, PSMColumn.MODIFICATIONS);
AddStableColumn(factory, PSMColumn.RETENTION_TIME);
AddStableColumn(factory, PSMColumn.CHARGE);
AddStableColumn(factory, PSMColumn.EXP_MASS_TO_CHARGE);
AddStableColumn(factory, PSMColumn.CALC_MASS_TO_CHARGE);
AddStableColumn(factory, PSMColumn.SPECTRA_REF);
AddStableColumn(factory, PSMColumn.PRE);
AddStableColumn(factory, PSMColumn.POST);
AddStableColumn(factory, PSMColumn.START);
AddStableColumn(factory, PSMColumn.END);
}
else if (section.Equals(Section.Small_Molecule_Header)){
AddStableColumn(factory, SmallMoleculeColumn.IDENTIFIER);
AddStableColumn(factory, SmallMoleculeColumn.CHEMICAL_FORMULA);
AddStableColumn(factory, SmallMoleculeColumn.SMILES);
AddStableColumn(factory, SmallMoleculeColumn.INCHI_KEY);
AddStableColumn(factory, SmallMoleculeColumn.DESCRIPTION);
AddStableColumn(factory, SmallMoleculeColumn.EXP_MASS_TO_CHARGE);
AddStableColumn(factory, SmallMoleculeColumn.CALC_MASS_TO_CHARGE);
AddStableColumn(factory, SmallMoleculeColumn.CHARGE);
AddStableColumn(factory, SmallMoleculeColumn.RETENTION_TIME);
AddStableColumn(factory, SmallMoleculeColumn.TAXID);
AddStableColumn(factory, SmallMoleculeColumn.SPECIES);
AddStableColumn(factory, SmallMoleculeColumn.DATABASE);
AddStableColumn(factory, SmallMoleculeColumn.DATABASE_VERSION);
AddStableColumn(factory, SmallMoleculeColumn.SPECTRA_REF);
AddStableColumn(factory, SmallMoleculeColumn.SEARCH_ENGINE);
AddStableColumn(factory, SmallMoleculeColumn.BEST_SEARCH_ENGINE_SCORE);
AddStableColumn(factory, SmallMoleculeColumn.MODIFICATIONS);
}
else{
throw new NotImplementedException("Section should be Protein, Peptide or " +
"Small_Molecule. Others can not setting. ");
}
foreach (var column in factory.StableColumnMapping){
factory.ColumnMapping.Add(column.Key, column.Value);
}
factory.section = section;
return factory;
}
/**
* valueLabel pattern like: column_name=param_string
*/
private MZTabError checkColUnit(string valueLabel, MZTabColumnFactory factory)
{
string[] items = valueLabel.Split('=');
string columnName = items[0].Trim();
string value = items[1].Trim();
MZTabColumn column = factory.FindColumnByHeader(columnName);
if (column == null){
// column_name not exists in the factory.
return new MZTabError(FormatErrorType.ColUnit, _lineNumber, valueLabel, columnName);
}
Param param = MZTabUtils.ParseParam(value);
if (param == null){
return new MZTabError(FormatErrorType.Param, _lineNumber, valueLabel, value);
}
if (factory.Section.Equals(Section.Protein_Header)){
_metadata.AddProteinColUnit(column, param);
}
else if (factory.Section.Equals(Section.Peptide_Header)){
_metadata.AddPeptideColUnit(column, param);
}
else if (factory.Section.Equals(Section.PSM_Header)){
_metadata.AddPSMColUnit(column, param);
}
else if (factory.Section.Equals(Section.Small_Molecule_Header)){
_metadata.AddSmallMoleculeColUnit(column, param);
}
return null;
}
public PSMLineParser(MZTabColumnFactory factory, PositionMapping positionMapping, Metadata metadata,
MZTabErrorList errorList)
: base(factory, positionMapping, metadata, errorList)
{
}
public Protein(MZTabColumnFactory factory)
: base(factory)
{
}