public NamedSequence Translate(bool dashAsMissing)
{
string aaSeq = Translate(Sequence, dashAsMissing);
NamedSequence namedAaSeq = new NamedSequence(Name, Protein, aaSeq);
return(namedAaSeq);
}
public override List <NamedSequence> Parse(TextReader reader)
{
if (NoHeader)
{
reader = new StringReader(OUTPUT_HEADER + "\n" + reader.ReadToEnd());
}
List <NamedSequence> seqs = new List <NamedSequence>();
string[] header = null;
foreach (Dictionary <string, string> row in SpecialFunctions.TabFileTable(reader, "", true, false, true))
{
if (header == null)
{
header = row[""].Split('\t');
Helper.CheckCondition(header.Length == 2, "Header does not conform to tab file format: " + row[""]);
continue;
}
string name = row[header[0]];
string seq = row[header[1]];
NamedSequence namedSeq = NamedSequence.Parse(name);
namedSeq.Sequence = RemoveWhiteSpace(seq);
seqs.Add(namedSeq);
}
return(seqs);
}
public override List <NamedSequence> Parse(TextReader reader)
{
List <NamedSequence> seqs = new List <NamedSequence>();
string line;
//string name = null;
NamedSequence seqToAdd = null;
StringBuilder sequence = new StringBuilder();
while ((line = reader.ReadLine()) != null)
{
if (line.StartsWith(">"))
{
if (seqToAdd != null)
{
//seqs.Add(new NamedSequence(name, RemoveWhiteSpace(sequence.ToString())));
seqToAdd.Sequence = sequence.ToString();
seqs.Add(seqToAdd);
}
seqToAdd = NamedSequence.Parse(line.Substring(1).Trim());
//name = line.Substring(1).Trim();
sequence = new StringBuilder(sequence.Length);
}
else
{
sequence.Append(line.Trim());
}
}
//seqs.Add(new NamedSequence(name, RemoveWhiteSpace(sequence.ToString())));
seqToAdd.Sequence = sequence.ToString();
seqs.Add(seqToAdd);
return(seqs);
}
public static NamedSequence Parse(string constructorArgs)
{
ConstructorArguments constArgs = new ConstructorArguments("(" + constructorArgs + ")");
NamedSequence result = new NamedSequence();
result.Protein = constArgs.ExtractOptional <string>("Protein", null);
result.Name = constArgs.ExtractNext <string>("name");
return(result);
}
public override List <NamedSequence> Parse(TextReader reader)
{
string line;
string[] fields;
line = reader.ReadLine().Trim();
if (!ConformsToFileFormat(line))
{
throw new FormatException("File does not conform to Phylip format. First line must be two integers specifying number of sequences and sequence length.");
}
fields = Regex.Split(line, @"\s+");
int lineCount, lineLength;
try
{
lineCount = int.Parse(fields[0]);
lineLength = int.Parse(fields[1]);
}
catch (FormatException e)
{
Console.WriteLine("Error parsing Phylip header: {0} split to {1} and {2}", line, fields[0], fields[1]);
throw e;
}
List <NamedSequence> seqs = new List <NamedSequence>(lineCount);
while ((line = reader.ReadLine()) != null)
{
if (!string.IsNullOrWhiteSpace(line))
{
int endOfName = Math.Max(10, line.IndexOfAny(new char[] { ' ', '\t' }, 0, Math.Min(50, line.Length)));
string name = line.Substring(0, endOfName).Trim();
string seq = RemoveWhiteSpace(line.Substring(endOfName));
//NamedSequence sequence = new NamedSequence(name, seq);
NamedSequence sequence = NamedSequence.Parse(name);
sequence.Sequence = seq;
if (sequence.Sequence.Length != lineLength)
{
throw new FormatException(string.Format("Expected {0} to be {1} chars long, but its {2} long.",
sequence.Name, lineLength, sequence.Sequence.Length));
}
seqs.Add(sequence);
}
}
if (seqs.Count != lineCount)
{
throw new FormatException(string.Format("Expected {0} sequences. Read {1}.", lineCount, seqs.Count));
}
return(seqs);
}
public override bool Equals(object obj)
{
NamedSequence other = obj as NamedSequence;
if (other == null)
{
return(false);
}
return(other.Name == this.Name && other.Sequence == this.Sequence);
}
public static NamedSequence GetConsensus(List <NamedSequence> seqs)
{
int len = -1;
//StringBuilder consensusSeq = new StringBuilder();
List <AASeq> aaSeqs = new List <AASeq>();
bool isDna = seqs[0].IsDna();
foreach (NamedSequence seq in seqs)
{
AASeq aaSeq = isDna ? DnaSeq.GetInstance(seq.Sequence, MixtureSemantics.Uncertainty) : AASeq.GetInstance(seq.Sequence, MixtureSemantics.Uncertainty);
len = Math.Max(len, aaSeq.Count);
//if (len < 0)
//{
// len = aaSeq.Count;
//}
//else
//{
// Helper.CheckCondition(len == aaSeq.Count, String.Format("Sequence {0} is a different length from previous sequences", seq.Name));
//}
aaSeqs.Add(aaSeq);
}
//for (int i = 0; i < len; i++)
//{
// Dictionary<string, int> charToCount = new Dictionary<string, int>();
// KeyValuePair<string, int> currentConsensus = new KeyValuePair<string, int>("z", -1);
// foreach (AASeq aaSeq in aaSeqs)
// {
// if (i < aaSeq.Count)
// {
// string residue = aaSeq.SubSeqAA0Pos(i, 1).ToString();
// charToCount[residue] = SpecialFunctions.GetValueOrDefault(charToCount, residue) + 1;
// if (charToCount[residue] > currentConsensus.Value)
// {
// currentConsensus = new KeyValuePair<string, int>(residue, charToCount[residue]);
// }
// }
// }
// consensusSeq.Append(currentConsensus.Key);
//}
AASeq consensusAaSeq = GetAaSeqConsensus(aaSeqs);
NamedSequence consensus = new NamedSequence("consensus", consensusAaSeq.ToString());
return(consensus);
}
/// <param name="readingFrameToTranslate">Will start translation at start+readingFrameToTranslate-1. If RF>3, will take the reverse
/// commplement, then translate the revComp using rf -= 3. The sequence will be padded at the end to make it the same length as the
/// original, and the final result will be reversed, so that the sequence is read in the same order as the original string!
/// ie RF is 1-based.</param>
public static List <NamedSequence> Translate(List <NamedSequence> seqs, bool nucToAaDashAsMissing = true, int readingFrameToTranslate = 1)
{
Helper.CheckCondition(readingFrameToTranslate > 0 && readingFrameToTranslate <= 6, "readingFrameToTranslate must be between 0 and 6. " + readingFrameToTranslate + " is not valid.");
bool isAntisense = false;
if (readingFrameToTranslate > 3)
{
seqs = NucSeqReverseComplement(seqs);
readingFrameToTranslate -= 3;
isAntisense = true;
}
List <NamedSequence> result = new List <NamedSequence>(seqs.Count);
int lastLen = -1;
foreach (NamedSequence seq in seqs)
{
string seqToTranslate = seq.Sequence.Substring(readingFrameToTranslate - 1);
if (seqToTranslate.Length != lastLen)
{
lastLen = seqToTranslate.Length;
if (lastLen % 3 != 0)
{
Console.Error.WriteLine("WARNING: Sequence of length {0} is not divisible by 3. Cutting off the end for translation.", lastLen);
}
}
if (seqToTranslate.Length % 3 != 0)
{
seqToTranslate = seqToTranslate.Substring(0, seqToTranslate.Length - seqToTranslate.Length % 3);
}
//seqToTranslate += Enumerable.Repeat("-", 3 - (seqToTranslate.Length % 3)).StringJoin();
var translatedSeq = new NamedSequence(seq.Name, seq.Protein, NamedSequence.Translate(seqToTranslate, nucToAaDashAsMissing));
result.Add(isAntisense ? translatedSeq.ReverseSequence() : translatedSeq);
}
return(result);
}
