/// <summary>
/// Returns standard feature class instance, if the key in the specified feature item is found
/// in the map; otherwise returns the specified feature item itself.
/// For example:
/// If the specified feature item has the key "Gene" then this method returns instance of the Gene class
/// with data copied from the specified item.
/// </summary>
/// <param name="item">Feature item instance to which the standard feature item instance is needed.</param>
/// <returns>If found returns appropriate class instance for the specified feature item, otherwise returns
/// the specified item itself.</returns>
public static FeatureItem GetStandardFeatureItem(FeatureItem item)
{
if (item == null)
{
throw new ArgumentNullException("item");
}
Type type = null;
if (featureMap.ContainsKey(item.Key))
{
type = featureMap[item.Key];
}
if (type != null)
{
FeatureItem newItem = (FeatureItem)Activator.CreateInstance(type, item.Location);
foreach (KeyValuePair<string, List<string>> kvp in item.Qualifiers)
{
newItem.Qualifiers.Add(kvp.Key, kvp.Value);
}
item = newItem;
}
return item;
}
/// <summary>
/// Private Constructor for clone method.
/// </summary>
/// <param name="other">FeatureItem instance to clone.</param>
protected FeatureItem(FeatureItem other)
{
Key = other.Key;
this.Location = other.Location.Clone();
Qualifiers = new Dictionary<string, List<string>>();
foreach (KeyValuePair<string, List<string>> kvp in other.Qualifiers)
{
if (kvp.Value != null)
{
Qualifiers.Add(kvp.Key, new List<string>(kvp.Value));
}
else
{
Qualifiers.Add(kvp.Key, null);
}
}
}
/// <summary>
/// Returns standard feature class instance, if the key in the specified feature item is found
/// in the map; otherwise returns the specified feature item itself.
/// For example:
/// If the specified feature item has the key "Gene" then this method returns instance of the Gene class
/// with data copied from the specified item.
/// </summary>
/// <param name="item">Feature item instance to which the standard feature item instance is needed.</param>
/// <returns>If found returns appropriate class instance for the specified feature item, otherwise returns
/// the specified item itself.</returns>
public static FeatureItem GetStandardFeatureItem(FeatureItem item)
{
Type type = null;
if (featureMap.ContainsKey(item.Key))
{
type = featureMap[item.Key];
}
if (type != null)
{
FeatureItem newItem = (FeatureItem)Activator.CreateInstance(type, item.Location);
foreach (KeyValuePair <string, List <string> > kvp in item.Qualifiers)
{
newItem.Qualifiers.Add(kvp.Key, kvp.Value);
}
item = newItem;
}
return(item);
}
/// <summary>
/// Helper method to parse the feature's qualifiers of gen bank data.
/// </summary>
/// <param name="metadata">feature object</param>
/// <param name="cellRange">Range of cells</param>
/// <param name="rowIndex">Current index of row</param>
/// <returns>Index of row</returns>
private static int ParseQualifiers(FeatureItem featureItem, object[,] cellRange, int rowIndex)
{
string message = string.Empty;
string key;
string subKey;
string value = string.Empty;
while (rowIndex < cellRange.GetLength(0))
{
if (3 > cellRange.GetLength(1))
{
message = String.Format(
CultureInfo.InvariantCulture,
Resources.UnrecognizedGenBankMetadataFormat,
REFERENCE);
throw new FormatException(message);
}
if (null != cellRange[rowIndex, KeyColumnIndex])
{
key = cellRange[rowIndex, KeyColumnIndex].ToString().ToUpperInvariant();
if (!string.IsNullOrWhiteSpace(key))
{
break;
}
}
subKey = cellRange[rowIndex, SubKeyColumnIndex] != null ? cellRange[rowIndex, SubKeyColumnIndex].ToString() : string.Empty;
if (string.IsNullOrWhiteSpace(subKey))
{
rowIndex++;
continue;
}
value = cellRange[rowIndex, ValueColumnIndex] != null ? cellRange[rowIndex, ValueColumnIndex].ToString() : string.Empty;
if (!featureItem.Qualifiers.ContainsKey(subKey))
{
featureItem.Qualifiers[subKey] = new List<string>();
}
featureItem.Qualifiers[subKey].Add(value);
rowIndex++;
}
return rowIndex;
}
/// <summary>
/// Parses the GenBank features from the GenBank file.
/// </summary>
/// <param name="line">parse line</param>
/// <param name="sequence">The sequence.</param>
/// <param name="stream">The stream reader.</param>
/// <returns>The parsed line.</returns>
private string ParseFeatures(string line, ref Sequence sequence, StreamReader stream)
{
ILocationBuilder locBuilder = LocationBuilder;
if (locBuilder == null)
{
throw new InvalidOperationException(Properties.Resource.NullLocationBuild);
}
// set data indent for features
string lineData;
// The sub-items of a feature are referred to as qualifiers. These do not have unique
// keys, so they are stored as lists in the SubItems dictionary.
SequenceFeatures features = new SequenceFeatures();
IList <FeatureItem> featureList = features.All;
while (line != null)
{
string lineHeader = GetLineHeader(line, FeatureDataIndent);
if (String.IsNullOrEmpty(line) || lineHeader == "FEATURES")
{
line = GoToNextLine(line, stream);
continue;
}
if (line[0] != ' ')
{
// start of non-feature text
break;
}
if (lineHeader == null)
{
string message = Properties.Resource.GenbankEmptyFeature;
Trace.Report(message);
throw new InvalidDataException(message);
}
// check for multi-line location string
lineData = GetLineData(line, FeatureDataIndent);
string featureKey = lineHeader;
string location = lineData;
line = GoToNextLine(line, stream);
lineData = GetLineData(line, FeatureDataIndent);
lineHeader = GetLineHeader(line, FeatureDataIndent);
while ((line != null) && (lineHeader == string.Empty) &&
(lineData != string.Empty) && !lineData.StartsWith("/", StringComparison.Ordinal))
{
location += lineData;
GetLineData(line, FeatureDataIndent);
line = GoToNextLine(line, stream);
lineData = GetLineData(line, FeatureDataIndent);
lineHeader = GetLineHeader(line, FeatureDataIndent);
}
// create features as MetadataListItems
FeatureItem feature = new FeatureItem(featureKey, locBuilder.GetLocation(location));
// process the list of qualifiers, which are each in the form of
// /key="value"
string qualifierKey = string.Empty;
string qualifierValue = string.Empty;
bool quotationMarkStarted = false;
while (line != null)
{
lineData = GetLineData(line, FeatureDataIndent);
lineHeader = GetLineHeader(line, FeatureDataIndent);
if ((lineHeader == string.Empty) && (lineData != null))
{
// '/' denotes a continuation of the previous line
// Note that, if there are multiple lines of qualifierValue,
// sometimes a line break will happen such that a "/" which is
// part of the qualifierValue will start a continuation line.
// This is identified by verifying open and closing double quotes.
if (lineData.StartsWith("/", StringComparison.Ordinal) && !quotationMarkStarted)
{
// new qualifier; save previous if this isn't the first
if (!String.IsNullOrEmpty(qualifierKey))
{
AddQualifierToFeature(feature, qualifierKey, qualifierValue);
}
// set the key and value of this qualifier
int equalsIndex = lineData.IndexOf('=');
if (equalsIndex < 0)
{
// no value, just key (this is allowed, see NC_005213.gbk)
qualifierKey = lineData.Substring(1);
qualifierValue = string.Empty;
}
else if (equalsIndex > 0)
{
qualifierKey = lineData.Substring(1, equalsIndex - 1);
qualifierValue = lineData.Substring(equalsIndex + 1);
quotationMarkStarted = qualifierValue[0] == '"';
if (qualifierValue[qualifierValue.Length - 1] == '"')
{
quotationMarkStarted = false;
}
}
else
{
string message = String.Format(
CultureInfo.CurrentCulture,
Properties.Resource.GenbankInvalidFeature,
line);
Trace.Report(message);
throw new InvalidDataException(message);
}
}
else
{
// Continuation of previous line; "note" gets a line break, and
// everything else except "translation" and "transl_except" gets a
// space to separate words.
if (qualifierKey == "note")
{
qualifierValue += Environment.NewLine;
}
else if (qualifierKey != "translation" && qualifierKey != "transl_except")
{
qualifierValue += " ";
}
qualifierValue += lineData;
if (qualifierValue[qualifierValue.Length - 1] == '"')
{
quotationMarkStarted = false;
}
}
line = GoToNextLine(line, stream);
}
else if (line.StartsWith("\t", StringComparison.Ordinal))
{
// this seems to be data corruption; but BioPerl test set includes
// (old, 2003) NT_021877.gbk which has this problem, so we
// handle it
ApplicationLog.WriteLine("WARN: nonstandard line format at line {0}: '{1}'", lineNumber, line);
qualifierValue += " " + line.Trim();
if (qualifierValue[qualifierValue.Length - 1] == '"')
{
quotationMarkStarted = false;
}
line = GoToNextLine(line, stream);
}
else
{
break;
}
}
// add last qualifier
if (!String.IsNullOrEmpty(qualifierKey))
{
AddQualifierToFeature(feature, qualifierKey, qualifierValue);
}
// still add feature, even if it has no qualifiers
featureList.Add(StandardFeatureMap.GetStandardFeatureItem(feature));
}
if (featureList.Count > 0)
{
((GenBankMetadata)sequence.Metadata[Helper.GenBankMetadataKey]).Features = features;
}
return(line);
}
/// <summary>
/// Validate addition of GenBank features.
/// </summary>
/// <param name="nodeName">xml node name.</param>
private void ValidateAdditionGenBankFeatures(string nodeName)
{
// Get Values from XML node.
string filePath = utilityObj.xmlUtil.GetTextValue(
nodeName, Constants.FilePathNode);
string alphabetName = utilityObj.xmlUtil.GetTextValue(
nodeName, Constants.AlphabetNameNode);
string expectedSequence = utilityObj.xmlUtil.GetTextValue(
nodeName, Constants.ExpectedSequenceNode);
string addFirstKey = utilityObj.xmlUtil.GetTextValue(
nodeName, Constants.FirstKey);
string addSecondKey = utilityObj.xmlUtil.GetTextValue(
nodeName, Constants.SecondKey);
string addFirstLocation = utilityObj.xmlUtil.GetTextValue(
nodeName, Constants.FirstLocation);
string addSecondLocation = utilityObj.xmlUtil.GetTextValue(
nodeName, Constants.SecondLocation);
string addFirstQualifier = utilityObj.xmlUtil.GetTextValue(
nodeName, Constants.FirstQualifier);
string addSecondQualifier = utilityObj.xmlUtil.GetTextValue(
nodeName, Constants.SecondQualifier);
ISequenceParser parser1 = new GenBankParser();
{
IEnumerable<ISequence> seqList1 = parser1.Parse(filePath);
var localBuilderObj = new LocationBuilder();
string tempFileName = Path.GetTempFileName();
string expectedUpdatedSequence =
expectedSequence.Replace("\r", "").Replace("\n", "").Replace(" ", "");
var orgSeq = new Sequence(Utility.GetAlphabet(alphabetName),
expectedUpdatedSequence);
orgSeq.ID = seqList1.ElementAt(0).ID;
orgSeq.Metadata.Add(Constants.GenBank,
seqList1.ElementAt(0).Metadata[Constants.GenBank]);
ISequenceFormatter formatterObj = new GenBankFormatter();
{
formatterObj.Format(orgSeq, tempFileName);
// parse GenBank file.
var parserObj = new GenBankParser();
{
IEnumerable<ISequence> seqList = parserObj.Parse(tempFileName);
ISequence seq = seqList.ElementAt(0);
var metadata = (GenBankMetadata) seq.Metadata[Constants.GenBank];
// Add a new features to Genbank features list.
metadata.Features = new SequenceFeatures();
var feature = new FeatureItem(addFirstKey, addFirstLocation);
var qualifierValues = new List<string>();
qualifierValues.Add(addFirstQualifier);
qualifierValues.Add(addFirstQualifier);
feature.Qualifiers.Add(addFirstQualifier, qualifierValues);
metadata.Features.All.Add(feature);
feature = new FeatureItem(addSecondKey, addSecondLocation);
qualifierValues = new List<string>();
qualifierValues.Add(addSecondQualifier);
qualifierValues.Add(addSecondQualifier);
feature.Qualifiers.Add(addSecondQualifier, qualifierValues);
metadata.Features.All.Add(feature);
// Validate added GenBank features.
Assert.AreEqual(metadata.Features.All[0].Key.ToString(null), addFirstKey);
Assert.AreEqual(
localBuilderObj.GetLocationString(metadata.Features.All[0].Location),
addFirstLocation);
Assert.AreEqual(metadata.Features.All[1].Key.ToString(null), addSecondKey);
Assert.AreEqual(localBuilderObj.GetLocationString(metadata.Features.All[1].Location),
addSecondLocation);
parserObj.Close();
}
File.Delete(tempFileName);
}
}
}
/// <summary>
/// Validate addition of GenBank features.
/// </summary>
/// <param name="nodeName">xml node name.</param>
private void ValidateGenBankSubFeatures(string nodeName)
{
// Get Values from XML node.
string firstKey = utilityObj.xmlUtil.GetTextValue(
nodeName, Constants.FirstKey);
string secondKey = utilityObj.xmlUtil.GetTextValue(
nodeName, Constants.SecondKey);
string thirdKey = utilityObj.xmlUtil.GetTextValue(
nodeName, Constants.ThirdFeatureKey);
string fourthKey = utilityObj.xmlUtil.GetTextValue(
nodeName, Constants.FourthKey);
string fifthKey = utilityObj.xmlUtil.GetTextValue(
nodeName, Constants.FifthKey);
string firstLocation = utilityObj.xmlUtil.GetTextValue(
nodeName, Constants.FirstLocation);
string secondLocation = utilityObj.xmlUtil.GetTextValue(
nodeName, Constants.SecondLocation);
string thirdLocation = utilityObj.xmlUtil.GetTextValue(
nodeName, Constants.ThirdLocation);
string fourthLocation = utilityObj.xmlUtil.GetTextValue(
nodeName, Constants.FourthLocation);
string fifthLocation = utilityObj.xmlUtil.GetTextValue(
nodeName, Constants.FifthLocation);
string featuresCount = utilityObj.xmlUtil.GetTextValue(
nodeName, Constants.MainFeaturesCount);
string secondCount = utilityObj.xmlUtil.GetTextValue(
nodeName, Constants.SecondCount);
string thirdCount = utilityObj.xmlUtil.GetTextValue(
nodeName, Constants.ThirdCount);
// Create a feature items
var seqFeatures = new SequenceFeatures();
var firstItem = new FeatureItem(firstKey, firstLocation);
var secondItem = new FeatureItem(secondKey, secondLocation);
var thirdItem = new FeatureItem(thirdKey, thirdLocation);
var fourthItem = new FeatureItem(fourthKey, fourthLocation);
var fifthItem = new FeatureItem(fifthKey, fifthLocation);
seqFeatures.All.Add(firstItem);
seqFeatures.All.Add(secondItem);
seqFeatures.All.Add(thirdItem);
seqFeatures.All.Add(fourthItem);
seqFeatures.All.Add(fifthItem);
// Validate sub features .
List<FeatureItem> subFeatures = firstItem.GetSubFeatures(seqFeatures);
Assert.AreEqual(Convert.ToInt32(featuresCount, null), subFeatures.Count);
subFeatures = secondItem.GetSubFeatures(seqFeatures);
Assert.AreEqual(Convert.ToInt32(secondCount, null), subFeatures.Count);
subFeatures = thirdItem.GetSubFeatures(seqFeatures);
Assert.AreEqual(Convert.ToInt32(thirdCount, null), subFeatures.Count);
}
/// <summary>
/// Parses the GenBank features from the GenBank file.
/// </summary>
/// <param name="line">parse line</param>
/// <param name="sequence">The sequence.</param>
/// <param name="stream">The stream reader.</param>
/// <returns>The parsed line.</returns>
private string ParseFeatures(string line, ref Sequence sequence, StreamReader stream)
{
ILocationBuilder locBuilder = LocationBuilder;
if (locBuilder == null)
{
throw new InvalidOperationException(Properties.Resource.NullLocationBuild);
}
// set data indent for features
string lineData;
// The sub-items of a feature are referred to as qualifiers. These do not have unique
// keys, so they are stored as lists in the SubItems dictionary.
SequenceFeatures features = new SequenceFeatures();
IList<FeatureItem> featureList = features.All;
while (line != null)
{
string lineHeader = GetLineHeader(line, FeatureDataIndent);
if (String.IsNullOrEmpty(line) || lineHeader == "FEATURES")
{
line = GoToNextLine(line, stream);
continue;
}
if (line[0] != ' ')
{
// start of non-feature text
break;
}
if (lineHeader == null)
{
string message = Properties.Resource.GenbankEmptyFeature;
Trace.Report(message);
throw new InvalidDataException(message);
}
// check for multi-line location string
lineData = GetLineData(line, FeatureDataIndent);
string featureKey = lineHeader;
string location = lineData;
line = GoToNextLine(line, stream);
lineData = GetLineData(line, FeatureDataIndent);
lineHeader = GetLineHeader(line, FeatureDataIndent);
while ((line != null) && (lineHeader == string.Empty) &&
(lineData != string.Empty) && !lineData.StartsWith("/", StringComparison.Ordinal))
{
location += lineData;
GetLineData(line, FeatureDataIndent);
line = GoToNextLine(line, stream);
lineData = GetLineData(line, FeatureDataIndent);
lineHeader = GetLineHeader(line, FeatureDataIndent);
}
// create features as MetadataListItems
FeatureItem feature = new FeatureItem(featureKey, locBuilder.GetLocation(location));
// process the list of qualifiers, which are each in the form of
// /key="value"
string qualifierKey = string.Empty;
string qualifierValue = string.Empty;
bool quotationMarkStarted = false;
while (line != null)
{
lineData = GetLineData(line, FeatureDataIndent);
lineHeader = GetLineHeader(line, FeatureDataIndent);
if ((lineHeader == string.Empty) && (lineData != null))
{
// '/' denotes a continuation of the previous line
// Note that, if there are multiple lines of qualifierValue,
// sometimes a line break will happen such that a "/" which is
// part of the qualifierValue will start a continuation line.
// This is identified by verifying open and closing double quotes.
if (lineData.StartsWith("/", StringComparison.Ordinal) && !quotationMarkStarted)
{
// new qualifier; save previous if this isn't the first
if (!String.IsNullOrEmpty(qualifierKey))
{
AddQualifierToFeature(feature, qualifierKey, qualifierValue);
}
// set the key and value of this qualifier
int equalsIndex = lineData.IndexOf('=');
if (equalsIndex < 0)
{
// no value, just key (this is allowed, see NC_005213.gbk)
qualifierKey = lineData.Substring(1);
qualifierValue = string.Empty;
}
else if (equalsIndex > 0)
{
qualifierKey = lineData.Substring(1, equalsIndex - 1);
qualifierValue = lineData.Substring(equalsIndex + 1);
quotationMarkStarted = qualifierValue[0] == '"';
if (qualifierValue[qualifierValue.Length - 1] == '"')
{
quotationMarkStarted = false;
}
}
else
{
string message = String.Format(
CultureInfo.CurrentCulture,
Properties.Resource.GenbankInvalidFeature,
line);
Trace.Report(message);
throw new InvalidDataException(message);
}
}
else
{
// Continuation of previous line; "note" gets a line break, and
// everything else except "translation" and "transl_except" gets a
// space to separate words.
if (qualifierKey == "note")
{
qualifierValue += Environment.NewLine;
}
else if (qualifierKey != "translation" && qualifierKey != "transl_except")
{
qualifierValue += " ";
}
qualifierValue += lineData;
if (qualifierValue[qualifierValue.Length - 1] == '"')
{
quotationMarkStarted = false;
}
}
line = GoToNextLine(line, stream);
}
else if (line.StartsWith("\t", StringComparison.Ordinal))
{
// this seems to be data corruption; but BioPerl test set includes
// (old, 2003) NT_021877.gbk which has this problem, so we
// handle it
ApplicationLog.WriteLine("WARN: nonstandard line format at line {0}: '{1}'", lineNumber, line);
qualifierValue += " " + line.Trim();
if (qualifierValue[qualifierValue.Length - 1] == '"')
{
quotationMarkStarted = false;
}
line = GoToNextLine(line, stream);
}
else
{
break;
}
}
// add last qualifier
if (!String.IsNullOrEmpty(qualifierKey))
{
AddQualifierToFeature(feature, qualifierKey, qualifierValue);
}
// still add feature, even if it has no qualifiers
featureList.Add(StandardFeatureMap.GetStandardFeatureItem(feature));
}
if (featureList.Count > 0)
{
((GenBankMetadata)sequence.Metadata[Helper.GenBankMetadataKey]).Features = features;
}
return line;
}
/// <summary>
/// Adds a qualifier to the feature object. The sub-items of a feature are referred to as qualifiers. These do not have unique
/// keys, so they are stored as lists in the SubItems dictionary.
/// </summary>
/// <param name="feature">The feature to which qualifier is to be added.</param>
/// <param name="qualifierKey">The qualifier key to be added to the feature.</param>
/// <param name="qualifierValue">The qualifier value.</param>
private static void AddQualifierToFeature(FeatureItem feature, string qualifierKey, string qualifierValue)
{
if (!feature.Qualifiers.ContainsKey(qualifierKey))
{
feature.Qualifiers[qualifierKey] = new List<string>();
}
feature.Qualifiers[qualifierKey].Add(qualifierValue);
}
/// <summary>
/// Returns list of citation references in this metadata which are referred in the specified feature.
/// </summary>
/// <param name="item">Feature Item.</param>
public List<CitationReference> GetCitationsReferredInFeature(FeatureItem item)
{
List<CitationReference> list = new List<CitationReference>();
if (item == null || !item.Qualifiers.ContainsKey(StandardQualifierNames.Citation))
{
return list;
}
foreach (string str in item.Qualifiers[StandardQualifierNames.Citation])
{
if (!string.IsNullOrEmpty(str))
{
string strCitationNumber = str.Replace("[", string.Empty).Replace("]", string.Empty);
int citationNumber = -1;
if (int.TryParse(strCitationNumber, out citationNumber))
{
CitationReference citation = References.FirstOrDefault(F => F.Number == citationNumber);
if (citation != null && !list.Contains(citation))
{
list.Add(citation);
}
}
}
}
return list;
}
private void ParseFeatures(BioTextReader bioReader, ref Sequence sequence)
{
ILocationBuilder locBuilder = LocationBuilder;
if (locBuilder == null)
{
throw new InvalidOperationException(Resource.NullLocationBuild);
}
// set data indent for features
bioReader.DataIndent = _featureDataIndent;
// The sub-items of a feature are referred to as qualifiers. These do not have unique
// keys, so they are stored as lists in the SubItems dictionary.
SequenceFeatures features = new SequenceFeatures();
IList <FeatureItem> featureList = features.All;
while (bioReader.HasLines)
{
if (String.IsNullOrEmpty(bioReader.Line) || bioReader.LineHeader == "FEATURES")
{
bioReader.GoToNextLine();
continue;
}
if (bioReader.Line[0] != ' ')
{
// start of non-feature text
break;
}
if (!bioReader.LineHasHeader)
{
string message = Properties.Resource.GenbankEmptyFeature;
Trace.Report(message);
throw new InvalidDataException(message);
}
// check for multi-line location string
string featureKey = bioReader.LineHeader;
string location = bioReader.LineData;
bioReader.GoToNextLine();
while (bioReader.HasLines && !bioReader.LineHasHeader &&
bioReader.LineHasData && !bioReader.LineData.StartsWith("/", StringComparison.Ordinal))
{
location += bioReader.LineData;
bioReader.GoToNextLine();
}
// create features as MetadataListItems
FeatureItem feature = new FeatureItem(featureKey, locBuilder.GetLocation(location));
// process the list of qualifiers, which are each in the form of
// /key="value"
string qualifierKey = string.Empty;
string qualifierValue = string.Empty;
while (bioReader.HasLines)
{
if (!bioReader.LineHasHeader && bioReader.LineHasData)
{
// '/' denotes a continuation of the previous line
if (bioReader.LineData.StartsWith("/", StringComparison.Ordinal))
{
// new qualifier; save previous if this isn't the first
if (!String.IsNullOrEmpty(qualifierKey))
{
AddQualifierToFeature(feature, qualifierKey, qualifierValue);
}
// set the key and value of this qualifier
int equalsIndex = bioReader.LineData.IndexOf('=');
if (equalsIndex < 0)
{
// no value, just key (this is allowed, see NC_005213.gbk)
qualifierKey = bioReader.LineData.Substring(1);
qualifierValue = string.Empty;
}
else if (equalsIndex > 0)
{
qualifierKey = bioReader.LineData.Substring(1, equalsIndex - 1);
qualifierValue = bioReader.LineData.Substring(equalsIndex + 1);
}
else
{
string message = String.Format(
CultureInfo.CurrentCulture,
Properties.Resource.GenbankInvalidFeature,
bioReader.Line);
Trace.Report(message);
throw new InvalidDataException(message);
}
}
else
{
// Continuation of previous line; "note" gets a line break, and
// everything else except "translation" and "transl_except" gets a
// space to separate words.
if (qualifierKey == "note")
{
qualifierValue += Environment.NewLine;
}
else if (qualifierKey != "translation" && qualifierKey != "transl_except")
{
qualifierValue += " ";
}
qualifierValue += bioReader.LineData;
}
bioReader.GoToNextLine();
}
else if (bioReader.Line.StartsWith("\t", StringComparison.Ordinal))
{
// this seems to be data corruption; but BioPerl test set includes
// (old, 2003) NT_021877.gbk which has this problem, so we
// handle it
ApplicationLog.WriteLine("WARN: nonstandard line format at line {0}: '{1}'",
bioReader.LineNumber, bioReader.Line);
qualifierValue += " " + bioReader.Line.Trim();
bioReader.GoToNextLine();
}
else
{
break;
}
}
// add last qualifier
if (!String.IsNullOrEmpty(qualifierKey))
{
AddQualifierToFeature(feature, qualifierKey, qualifierValue);
}
// still add feature, even if it has no qualifiers
featureList.Add(StandardFeatureMap.GetStandardFeatureItem(feature));
}
if (featureList.Count > 0)
{
((GenBankMetadata)sequence.Metadata[Helper.GenBankMetadataKey]).Features = features;
}
}
public void ValidateAdditionSingleGenBankQualifier()
{
// Get Values from XML node.
string filePath = utilityObj.xmlUtil.GetTextValue(
Constants.MediumSizeRNAGenBankFeaturesNode, Constants.FilePathNode);
string addFirstKey = utilityObj.xmlUtil.GetTextValue(
Constants.MediumSizeRNAGenBankFeaturesNode, Constants.FirstKey);
string addFirstLocation = utilityObj.xmlUtil.GetTextValue(
Constants.MediumSizeRNAGenBankFeaturesNode, Constants.FirstLocation);
string addFirstQualifier = utilityObj.xmlUtil.GetTextValue(
Constants.MediumSizeRNAGenBankFeaturesNode, Constants.FirstQualifier);
string addSecondQualifier = utilityObj.xmlUtil.GetTextValue(
Constants.MediumSizeRNAGenBankFeaturesNode, Constants.SecondQualifier);
// Parse a GenBank file.
ISequenceParser parserObj = new GenBankParser();
{
IEnumerable<ISequence> seqList = parserObj.Parse(filePath);
var locBuilder = new LocationBuilder();
// Validate Minus35Signal feature all qualifiers.
var metadata =
(GenBankMetadata) seqList.ElementAt(0).Metadata[Constants.GenBank];
// Add a new features to Genbank features list.
metadata.Features = new SequenceFeatures();
var feature = new FeatureItem(addFirstKey, addFirstLocation);
var qualifierValues = new List<string>();
qualifierValues.Add(addFirstQualifier);
qualifierValues.Add(addFirstQualifier);
feature.Qualifiers.Add(addFirstQualifier, qualifierValues);
metadata.Features.All.Add(feature);
qualifierValues = new List<string>();
qualifierValues.Add(addSecondQualifier);
qualifierValues.Add(addSecondQualifier);
feature.Qualifiers.Add(addSecondQualifier, qualifierValues);
metadata.Features.All.Add(feature);
// Validate added GenBank features.
Assert.AreEqual(
metadata.Features.All[0].Key.ToString(null), addFirstKey);
Assert.AreEqual(
locBuilder.GetLocationString(metadata.Features.All[0].Location),
addFirstLocation);
}
}
