public void TestReadFromFasta()
{
FastaFormat ff = new FastaFormat();
List <Sequence> seqs = SequenceUtils.Read(ff, @TestContext.CurrentContext.TestDirectory + "/../../../data//Standard_Protein_FIT_060222.noredundant.fasta");
Assert.AreEqual(43, seqs.Count);
}
/// <summary>
/// Fill reference allele from genome fasta file.
/// </summary>
/// <param name="snpItems"></param>
/// <param name="fastaFile"></param>
/// <param name="progress"></param>
public static void FillReferenceAlleleFromFasta(this IEnumerable<SNPItem> snpItems, string fastaFile, IProgressCallback progress = null)
{
if (progress == null)
{
progress = new ConsoleProgressCallback();
}
var dic = snpItems.ToGroupDictionary(m => m.Chrom);
progress.SetMessage("Filling reference allele from {0} file ...", fastaFile);
using (var sw = new StreamReader(fastaFile))
{
var ff = new FastaFormat();
Sequence seq;
while ((seq = ff.ReadSequence(sw)) != null)
{
progress.SetMessage("chromosome " + seq.Name + " ...");
var chr = HumanChromosomeToInt(seq.Name);
if (dic.ContainsKey(chr))
{
var snps = dic[chr];
foreach (var snp in snps)
{
snp.RefChar = char.ToUpper(seq.SeqString[snp.Position - 1]);
}
}
}
}
progress.SetMessage("Filling reference allele finished.");
}
public static Dictionary <char, double> GetDatabaseComposition(string fastaFile)
{
Dictionary <char, double> result = new Dictionary <char, double>();
using (StreamReader sr = new StreamReader(fastaFile))
{
FastaFormat ff = new FastaFormat();
Sequence seq;
while ((seq = ff.ReadSequence(sr)) != null)
{
foreach (char c in seq.SeqString)
{
if (result.ContainsKey(c))
{
result[c] = result[c] + 1;
}
else
{
result[c] = 1;
}
}
}
double total = result.Values.Sum();
foreach (char c in result.Keys.ToArray())
{
result[c] = result[c] / total;
}
}
return(result);
}
public void TestReadSequence()
{
FastaFormat ff = new FastaFormat();
StreamReader sr = new StreamReader("../../../data/test.fasta");
Sequence test1 = ff.ReadSequence(sr);
Assert.AreEqual("test1", test1.Name);
Assert.AreEqual("AAAAA", test1.SeqString);
Sequence test2 = ff.ReadSequence(sr);
Assert.AreEqual("test2", test2.Name);
Assert.AreEqual("BBBBB", test2.SeqString);
Sequence test3 = ff.ReadSequence(sr);
Assert.IsNull(test3);
}
public void TestReadSequence()
{
FastaFormat ff = new FastaFormat();
StreamReader sr = new StreamReader(TestContext.CurrentContext.TestDirectory + "/../../../data//test.fasta");
Sequence test1 = ff.ReadSequence(sr);
Assert.AreEqual("test1", test1.Name);
Assert.AreEqual("AAAAA", test1.SeqString);
Sequence test2 = ff.ReadSequence(sr);
Assert.AreEqual("test2", test2.Name);
Assert.AreEqual("BBBBB", test2.SeqString);
Sequence test3 = ff.ReadSequence(sr);
Assert.IsNull(test3);
}
public static Dictionary<string, Sequence> GetAccessNumberMap(string database, IStringParser<string> acParser)
{
Dictionary<string, Sequence> result = new Dictionary<string, Sequence>();
using (StreamReader sr = new StreamReader(database))
{
FastaFormat sf = new FastaFormat();
Sequence seq;
while ((seq = sf.ReadSequence(sr)) != null)
{
string ac = acParser.GetValue(seq.Name);
result[ac] = seq;
}
}
return result;
}
public static Dictionary <string, Sequence> GetAccessNumberMap(string database, IStringParser <string> acParser)
{
Dictionary <string, Sequence> result = new Dictionary <string, Sequence>();
using (StreamReader sr = new StreamReader(database))
{
FastaFormat sf = new FastaFormat();
Sequence seq;
while ((seq = sf.ReadSequence(sr)) != null)
{
string ac = acParser.GetValue(seq.Name);
result[ac] = seq;
}
}
return(result);
}
private void ProcessFile(ref int index, StreamWriter sw, string fastaFile, bool isContaminant)
{
FastaFormat ff = new FastaFormat();
using (StreamReader sr = new StreamReader(fastaFile))
{
Progress.SetRange(0, sr.BaseStream.Length);
Sequence seq;
while ((seq = ff.ReadSequence(sr)) != null)
{
Progress.SetPosition(StreamUtils.GetCharpos(sr));
if (isContaminant)
{
if (!seq.Reference.StartsWith("CON_"))
{
seq.Reference = "CON_" + seq.Reference;
}
}
if (combined)
{
ff.WriteSequence(sw, seq);
}
if (pseudoAminoacid)
{
builder.Build(seq);
}
index++;
Sequence reversedSeq = SequenceUtils.GetReversedSequence(seq.SeqString, index);
ff.WriteSequence(sw, reversedSeq);
}
}
}
public static IAccessNumberParser GuessParser(string databaseFileName)
{
FastaFormat ff = new FastaFormat();
using (StreamReader sr = new StreamReader(databaseFileName))
{
Sequence seq;
while ((seq = ff.ReadSequence(sr)) != null)
{
if (seq.Reference.StartsWith("IPI"))
{
return(new NoExceptionAccessNumberParser(new IPIAccessNumberParser()));
}
if (seq.Reference.StartsWith("gi|"))
{
return(new NoExceptionAccessNumberParser(new NRAccessNumberParser()));
}
}
}
return(DefaultAccessNumberParser.GetInstance());
}
public static IAccessNumberParser GuessParser(string databaseFileName)
{
FastaFormat ff = new FastaFormat();
using (StreamReader sr = new StreamReader(databaseFileName))
{
Sequence seq;
while ((seq = ff.ReadSequence(sr)) != null)
{
if (seq.Reference.StartsWith("IPI"))
{
return new NoExceptionAccessNumberParser(new IPIAccessNumberParser());
}
if (seq.Reference.StartsWith("gi|"))
{
return new NoExceptionAccessNumberParser(new NRAccessNumberParser());
}
}
}
return DefaultAccessNumberParser.GetInstance();
}
public static List<CoverageRegion> GetTargetCoverageRegion(ITargetBuilderOptions options, IProgressCallback progress, bool removeRegionWithoutSequence = true)
{
List<CoverageRegion> result;
if (options.TargetFile.EndsWith(".xml"))
{
result = GetTargetCoverageRegionFromXml(options, progress);
}
else
{
result = GetTargetCoverageRegionFromBed(options, progress);
}
var dic = result.ToGroupDictionary(m => m.Seqname);
progress.SetMessage("Filling sequence from {0}...", options.GenomeFastaFile);
using (var sr = new StreamReader(options.GenomeFastaFile))
{
var ff = new FastaFormat();
Sequence seq;
while ((seq = ff.ReadSequence(sr)) != null)
{
progress.SetMessage("Processing chromosome {0} ...", seq.Reference);
var seqname = seq.Name.StringAfter("chr");
List<CoverageRegion> lst;
if (dic.TryGetValue(seqname, out lst))
{
foreach (var l in lst)
{
l.Sequence = seq.SeqString.Substring((int)(l.Start - 1), (int)l.Length);
if(l.Strand == '+')
{
l.ReverseComplementedSequence = SequenceUtils.GetReverseComplementedSequence(l.Sequence);
}
}
}
}
}
if (removeRegionWithoutSequence)
{
result.RemoveAll(l => string.IsNullOrEmpty(l.Sequence));
}
progress.SetMessage("Filling sequence finished.");
var namemap = new MapReader(1, 12).ReadFromFile(options.RefgeneFile);
result.ForEach(m =>
{
var gene = m.Name.StringBefore("_utr3");
m.GeneSymbol = namemap.ContainsKey(gene) ? namemap[gene] : string.Empty;
});
return result;
}
public void TestReadFromFasta()
{
FastaFormat ff = new FastaFormat();
List<Sequence> seqs = SequenceUtils.Read(ff, @"../../../data/Standard_Protein_FIT_060222.noredundant.fasta");
Assert.AreEqual(43, seqs.Count);
}
public static Dictionary<char, double> GetDatabaseComposition(string fastaFile)
{
Dictionary<char, double> result = new Dictionary<char, double>();
using (StreamReader sr = new StreamReader(fastaFile))
{
FastaFormat ff = new FastaFormat();
Sequence seq;
while ((seq = ff.ReadSequence(sr)) != null)
{
foreach (char c in seq.SeqString)
{
if (result.ContainsKey(c))
{
result[c] = result[c] + 1;
}
else
{
result[c] = 1;
}
}
}
double total = result.Values.Sum();
foreach (char c in result.Keys.ToArray())
{
result[c] = result[c] / total;
}
}
return result;
}
