/// <summary>
/// Search sequence online in the corresponding DB
/// </summary>
public void RetrieveSequences()
{
WSDBFetchServerService db = new WSDBFetchServerService();
foreach (PSIPIanalysissearchDBSequenceType p in ListProteins)
{
if (p.seq != null)
{
continue;
}
if (p.SearchDatabase_ref.ToUpper() != "SDB_SWISSPROT")
{
throw new ApplicationException("DB " + p.SearchDatabase_ref + " not supported");
}
Console.WriteLine("Retrieving " + p.accession + " ...");
string fasta = db.fetchData("UNIPROT:" + p.accession, "fasta", "raw");
fasta = fasta.Remove(0, fasta.IndexOf('\n'));
string seq = "";
foreach (string str in fasta.Split('\n'))
{
seq += str;
}
p.seq = seq; //seq.ToUpper();
}
}
public string GetFASTA(string accessionNumber)
{
WSDBFetchServerService fetch = new WSDBFetchServerService();
try
{
string result = fetch.fetchData("UniProtKB: Accession: " + accessionNumber, "FASTA", "RAW");
return(result);
}
catch (Exception e)
{
string problem = "There was a Problem with your request. Please check that your accession number is correct and that the record is on the UNIprotKB database.";
return(problem);
}
}
public string[] GetDatabases()
{
string[] result = null;
WSDBFetchServerService fetch = new WSDBFetchServerService();
try
{
result = fetch.getSupportedDBs();
return(result);
}
catch (Exception e)
{
result[0] = "NO DATABASES FOUND";
}
return(result);
}
/// <summary>
/// Search sequence online in the corresponding DB
/// </summary>
public void RetrieveSequences()
{
WSDBFetchServerService db = new WSDBFetchServerService();
foreach( DBSequenceType p in ListProteins ) {
if( p.Seq!= null )
continue;
if( p.searchDatabase_ref.ToUpper() != "SDB_SWISSPROT" )
throw new ApplicationException( "DB " + p.searchDatabase_ref + " not supported" );
Console.WriteLine( "Retrieving " + p.accession + " ..." );
string fasta = db.fetchData("UNIPROT:"+p.accession, "fasta", "raw");
fasta = fasta.Remove(0,fasta.IndexOf('\n'));
string seq = "";
foreach( string str in fasta.Split('\n') )
seq += str;
p.Seq = seq;//seq.ToUpper();
}
}
public static void runCosmic( string file )
{
StreamReader rd = new StreamReader(new GZipStream(new FileStream(file,FileMode.Open), CompressionMode.Decompress));
string line;
char[] sep1 = new char[]{','};
char[] sep2 = new char[]{':'};
string[] fields, fields2;
String mut, prot;
Variant v;
Fasta f;
List<string> ids = new List<string>();
SortedList<string,Fasta> fasta = new SortedList<string, Fasta>();
bool first = true;
//int count = 1;
while( (line=rd.ReadLine()) != null ) {
if( first ) {
first = false;
continue;
}
fields = line.Split(sep1);
if( fields[4].Length == 0 )
continue;
if( fasta.ContainsKey(fields[4]) )
f = fasta[fields[4]];
else {
WSDBFetchServerService server = new WSDBFetchServerService();
server.
f = new Fasta(Fasta.Type.Protein,fields[4],seq);
}
v = new Variant();
if( ids.Contains(fields[4]) )
continue;
ids.Add(fields[0]);
sreq.id = fields[0];
sres = ssrv.run_eFetch( sreq );
if( sres == null || sres.ExchangeSet.Rs == null || sres.ExchangeSet.Rs.Length == 0 )
continue;
Console.WriteLine( fields[0] + "..." );
foreach( string str in sres.ExchangeSet.Rs[0].hgvs ) {
if( !str.StartsWith("NP_") )
continue;
Console.Write( str + " " );
v = new Variant();
v.id = str;
fields2 = str.Split(sep2);
mut = fields2[1]; prot = fields2[0];
try {
v.orig = AminoAcid.Get(mut.Substring(2,3)).Letter;
v.mut = AminoAcid.Get(mut.Substring(mut.Length-3,3)).Letter;
v.pos = ulong.Parse(mut.Substring(5,mut.Length-8))-1;
} catch {
Console.WriteLine( "(filtered)" );
continue;
}
if( fasta.ContainsKey(prot) ) {
Console.WriteLine( "(cached)" );
f = fasta[prot];
} else {
preq.db = "protein";
preq.id = prot;
pres = psrv.run_eFetch( preq );
f = new Fasta(Fasta.Type.Protein,prot+"|"+pres.GBSet[0].GBSeq_definition,pres.GBSet[0].GBSeq_sequence);
fasta.Add( prot, f );
Console.WriteLine( "(downloaded)" );
}
f.mVariants.Add( v );
}
/*if( --count == 0 )
break;*/
}
foreach( Fasta fas in fasta.Values )
fas.Dump( true );
}
public static void runCosmic(string file)
{
StreamReader rd = new StreamReader(new GZipStream(new FileStream(file, FileMode.Open), CompressionMode.Decompress));
string line;
char[] sep1 = new char[] { ',' };
char[] sep2 = new char[] { ':' };
string[] fields, fields2;
String mut, prot;
Variant v;
Fasta f;
List <string> ids = new List <string>();
SortedList <string, Fasta> fasta = new SortedList <string, Fasta>();
bool first = true;
//int count = 1;
while ((line = rd.ReadLine()) != null)
{
if (first)
{
first = false;
continue;
}
fields = line.Split(sep1);
if (fields[4].Length == 0)
{
continue;
}
if (fasta.ContainsKey(fields[4]))
{
f = fasta[fields[4]];
}
else
{
WSDBFetchServerService server = new WSDBFetchServerService();
server.
f = new Fasta(Fasta.Type.Protein, fields[4], seq);
}
v = new Variant();
if (ids.Contains(fields[4]))
{
continue;
}
ids.Add(fields[0]);
sreq.id = fields[0];
sres = ssrv.run_eFetch(sreq);
if (sres == null || sres.ExchangeSet.Rs == null || sres.ExchangeSet.Rs.Length == 0)
{
continue;
}
Console.WriteLine(fields[0] + "...");
foreach (string str in sres.ExchangeSet.Rs[0].hgvs)
{
if (!str.StartsWith("NP_"))
{
continue;
}
Console.Write(str + " ");
v = new Variant();
v.id = str;
fields2 = str.Split(sep2);
mut = fields2[1]; prot = fields2[0];
try {
v.orig = AminoAcid.Get(mut.Substring(2, 3)).Letter;
v.mut = AminoAcid.Get(mut.Substring(mut.Length - 3, 3)).Letter;
v.pos = ulong.Parse(mut.Substring(5, mut.Length - 8)) - 1;
} catch {
Console.WriteLine("(filtered)");
continue;
}
if (fasta.ContainsKey(prot))
{
Console.WriteLine("(cached)");
f = fasta[prot];
}
else
{
preq.db = "protein";
preq.id = prot;
pres = psrv.run_eFetch(preq);
f = new Fasta(Fasta.Type.Protein, prot + "|" + pres.GBSet[0].GBSeq_definition, pres.GBSet[0].GBSeq_sequence);
fasta.Add(prot, f);
Console.WriteLine("(downloaded)");
}
f.mVariants.Add(v);
}
/*if( --count == 0 )
* break;*/
}
foreach (Fasta fas in fasta.Values)
{
fas.Dump(true);
}
}
