/// <summary>
/// Reads the proteins from the FASTA file.
/// </summary>
private void LoadFastaFile()
{
this.FastaEntries.Clear();
if (!string.IsNullOrEmpty(this.FastaDbFilePath) && File.Exists(this.FastaDbFilePath))
{
try
{
this.FastaEntries.AddRange(FastaReaderWriter.ReadFastaFile(this.FastaDbFilePath));
}
catch (FormatException e)
{
this.dialogService.ExceptionAlert(e);
this.FastaEntries.Clear();
}
}
}
public void CountMissingIdsInFasta(string missingIdsFile, string fastaFile)
{
string directoryPath = @"C:\Users\wilk011\Documents\DataFiles\FlashUnitTest";
missingIdsFile = Path.Combine(directoryPath, missingIdsFile);
fastaFile = Path.Combine(directoryPath, fastaFile);
// Parse missing ids file
List <string[]> missingProteins = new List <string[]>(); // each element of the list corresponds to an ID. Each element of the array corresponds to a protein accession
foreach (var line in File.ReadLines(missingIdsFile))
{
var parts = line.Split('\t');
if (parts.Length < 5)
{
continue;
}
var proteins = parts[2].Split(';').Where(part => !string.IsNullOrWhiteSpace(part)).ToArray();
missingProteins.Add(proteins);
}
// Parse FASTA file
var fastas = FastaReaderWriter.ReadFastaFile(fastaFile).ToDictionary(f => f.ProteinName, f => f);
// Count number of proteins
int numInFasta = 0, numNotInFasta = 0;
foreach (var proteinSet in missingProteins)
{
if (proteinSet.Any(protein => fastas.ContainsKey(protein)))
{
numInFasta++;
}
else
{
numNotInFasta++;
}
}
Console.WriteLine("Total: " + missingProteins.Count);
Console.WriteLine("# In FASTA:" + numInFasta);
Console.WriteLine("# Not in FASTA: " + numNotInFasta);
}
public void CountFlashIdsInFasta(string uniqueIdFile, string fastaFile)
{
string directoryPath = @"C:\Users\wilk011\Documents\DataFiles\FlashUnitTest";
uniqueIdFile = Path.Combine(directoryPath, uniqueIdFile);
fastaFile = Path.Combine(directoryPath, fastaFile);
var sequences = File.ReadLines(uniqueIdFile).Select(line => line.Split('\t')).Select(p => p[1]).ToList();
var fastas = FastaReaderWriter.ReadFastaFile(fastaFile);
int numUniqueInFasta = 0;
foreach (var sequence in sequences)
{
if (fastas.Any(fasta => fasta.ProteinSequenceText.Contains(sequence)))
{
numUniqueInFasta++;
}
}
Console.WriteLine(@"{0} of {1} in {2}", numUniqueInFasta, sequences.Count, Path.GetFileNameWithoutExtension(fastaFile));
}
/// <summary>
/// Create a truncated FASTA file based on selected proteins.
/// </summary>
/// <returns>The path to the truncated FASTA database file.</returns>
private string CreateTruncatedFastaFile()
{
var fastaFileName = Path.GetFileNameWithoutExtension(this.FastaDbFilePath);
var filePath = string.Format("{0}\\{1}_truncated.fasta", this.OutputFilePath, fastaFileName);
IEnumerable <FastaEntry> entries = new FastaEntry[0];
if (this.FromFastaEntry)
{
entries = this.FastaEntries.Where(entry => entry.Selected);
}
else if (this.FromSequence)
{
var selectedEntries = this.SequenceProteins.Where(entry => entry.Selected).ToArray();
if (this.FixedNTerm && this.FixedCTerm)
{ // Just use the selected sequence for every protein.
entries = selectedEntries.Select(
entry =>
new FastaEntry
{
ProteinName = entry.ProteinName,
ProteinDescription = entry.ProteinDescription,
ProteinSequenceText = this.SelectedSequence,
Selected = true
});
entries = new List <FastaEntry> {
entries.FirstOrDefault()
};
}
else if (this.FixedNTerm)
{
entries = from entry in selectedEntries
let startIndex = entry.ProteinSequenceText.IndexOf(this.SelectedSequence, StringComparison.Ordinal)
where startIndex > -1
let sequence = entry.ProteinSequenceText.Substring(startIndex)
select new FastaEntry
{
ProteinName = entry.ProteinName,
ProteinDescription = entry.ProteinDescription,
ProteinSequenceText = sequence
};
}
else if (this.FixedCTerm)
{
entries = from entry in selectedEntries
let startIndex = entry.ProteinSequenceText.IndexOf(this.SelectedSequence, StringComparison.Ordinal)
where startIndex > -1
let sequence = entry.ProteinSequenceText.Substring(0, startIndex + this.SelectedSequence.Length)
select new FastaEntry
{
ProteinName = entry.ProteinName,
ProteinDescription = entry.ProteinDescription,
ProteinSequenceText = sequence
};
}
else
{
entries = selectedEntries;
}
}
Console.WriteLine(@"Creating truncated fasta file at: {0}", filePath);
FastaReaderWriter.Write(entries, filePath);
return(filePath);
}