public static void MatchPeptideToProtein()
{
string pepSeq = "AAIKEESEGKLK";
string protID = "comp320_c0_seq1";//comp4598_c0_seq1";//comp1011_c0_seq1";
NucleicAcid.InitHash();
FileStream protein_fasta_database1 = new FileStream(@"G:\Thibault\Olivier\Databases\DMatton\Matton_Illumina_Anthesis_WithReverse.fasta", FileMode.Open, FileAccess.Read, FileShare.Read);
List <Protein> proteins1 = new List <Protein>(ProteinFastaReader.ReadProteins(protein_fasta_database1, false));
foreach (Protein prot in proteins1)
{
if (prot.Description.CompareTo(protID) == 0)
{
for (int shift = 0; shift < 3; shift++)
{
foreach (string seq in NucleicAcid.ConvertNA3ToAAs(prot.BaseSequence, shift, false))
{
if (seq.Contains("*") || seq.Contains("X"))
{
Console.WriteLine("AWRA");
}
if (seq.Contains(pepSeq))
{
Console.WriteLine("FORWARD (" + shift + ") : Found " + pepSeq + " in " + protID + " [" + seq + "]");
}
}
foreach (string seq in NucleicAcid.ConvertNA3ToAAs(prot.BaseSequence, shift, true))
{
if (seq.Contains("*") || seq.Contains("X"))
{
Console.WriteLine("AWRA");
}
if (seq.Contains(pepSeq))
{
Console.WriteLine("REVERSE (" + shift + ") : Found " + pepSeq + " in " + protID + " [" + seq + "]");
}
}
}
}
}
}
public static void YangLiuPeptidesWithAllProteins()
{
//vsCSV csvPeptides = new vsCSV(@"G:\Thibault\-=Proteomics_Raw_Data=-\ELITE\DEC18_2012\DMatton\Clustering_186716\Identifications.csv");
vsCSV csvPeptides = new vsCSV(@"G:\Thibault\-=Proteomics_Raw_Data=-\ELITE\DEC18_2012\DMatton\Clustering_186716\Cluster_Intensity_peptides_NormP.csv");
vsCSVWriter writer = new vsCSVWriter(@"G:\Thibault\-=Proteomics_Raw_Data=-\ELITE\DEC18_2012\DMatton\Clustering_186716\ProteinsPerPeptidesFromDatabases_AllReadingFrames.csv");
NucleicAcid.InitHash();
FileStream protein_fasta_database1 = new FileStream(@"G:\Thibault\Olivier\Databases\DMatton\Matton_Illumina_Anthesis_WithReverse.fasta", FileMode.Open, FileAccess.Read, FileShare.Read);
List <Protein> proteins1 = new List <Protein>(ProteinFastaReader.ReadProteins(protein_fasta_database1, false));
Dictionary <string, List <string> > protein1AAs = new Dictionary <string, List <string> >();
foreach (Protein prot in proteins1)
{
for (int shift = 0; shift < 3; shift++)
{
protein1AAs.Add(prot.Description + " | Reading Frame " + shift + " | Forward", NucleicAcid.ConvertNA3ToAAs(prot.BaseSequence, shift, false));
protein1AAs.Add(prot.Description + " | Reading Frame " + shift + " | Backward", NucleicAcid.ConvertNA3ToAAs(prot.BaseSequence, shift, true));
}
}
FileStream protein_fasta_database2 = new FileStream(@"G:\Thibault\Olivier\Databases\DMatton\mattond_20110418_WithReverse_EditedJuly2013.fasta", FileMode.Open, FileAccess.Read, FileShare.Read);
List <Protein> proteins2 = new List <Protein>(ProteinFastaReader.ReadProteins(protein_fasta_database2, false));
Dictionary <string, List <string> > protein2AAs = new Dictionary <string, List <string> >();
foreach (Protein prot in proteins2)
{
for (int shift = 0; shift < 3; shift++)
{
protein2AAs.Add(prot.Description + " | Reading Frame " + shift + " | Forward", NucleicAcid.ConvertNA3ToAAs(prot.BaseSequence, shift, false));
protein2AAs.Add(prot.Description + " | Reading Frame " + shift + " | Backward", NucleicAcid.ConvertNA3ToAAs(prot.BaseSequence, shift, true));
}
}
writer.AddLine(csvPeptides.LINES_LIST[0]);
Dictionary <string, List <string> > dicOfPepProt = new Dictionary <string, List <string> >();
for (int i = 1; i < csvPeptides.LINES_LIST.Count; i++)
{
string[] splits = csvPeptides.LINES_LIST[i].Split(vsCSV._Generic_Separator);
string seq = splits[4];
//string seq = splits[13];
/*
* string protDesc = splits[10];
* if (protein1AAs.ContainsKey(protDesc))
* if (!protein1AAs[protDesc].Contains(seq))
* Console.WriteLine("Should be there 1");
*
* if (protein2AAs.ContainsKey(protDesc))
* if (!protein2AAs[protDesc].Contains(seq))
* Console.WriteLine("Should be there 1");
* //*/
StringBuilder sb = new StringBuilder();
foreach (string key in protein1AAs.Keys)
{
foreach (string protSeq in protein1AAs[key])
{
if (protSeq.Contains(seq))
{
sb.Append(key + ";");
break;
}
}
}
foreach (string key in protein2AAs.Keys)
{
foreach (string protSeq in protein2AAs[key])
{
if (protSeq.Contains(seq))
{
sb.Append(key + ";");
break;
}
}
}
if (sb.Length == 0)
{
Console.WriteLine("Zut");
}
writer.AddLine(csvPeptides.LINES_LIST[i] + "," + sb.ToString().Trim());
}
writer.WriteToFile();
}