public double GetScore(WregexResult w)
{
if (m_pssm.Count == 0)
{
return(0.0);
}
double score = 0.0;
int i, j, k, len;
for (i = 0; i < w.Groups.Count; i++)
{
for (j = 0; j < m_pssm.Count && m_pssm[j].order != i; j++)
{
if (m_pssm[j].order >= w.Groups.Count)
{
throw new ApplicationException("Regex and PSSM have diferent lengths");
}
}
if (j == m_pssm.Count)
{
continue;
}
//Console.WriteLine( "DEBUG: " + w.Groups[i] + ", " + w.Groups[i].Length );
len = w.Groups[i].Length;
for (k = 0; k < len; k++)
{
score += GetScore(m_pssm[j], AminoAcid.Get((w.Groups[i])[k])); ///len;
}
}
return(Math.Pow(10, score / w.Match.Length) * 100);
}
public static void runEnsembl(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>();
eFetchSnpService ssrv = new eFetchSnpService();
EhuBio.Database.Ncbi.eFetch.Snp.MessageEFetchRequest sreq = new EhuBio.Database.Ncbi.eFetch.Snp.MessageEFetchRequest();
EhuBio.Database.Ncbi.eFetch.Snp.MessageEFetchResult sres;
eFetchSequenceService psrv = new eFetchSequenceService();
EhuBio.Database.Ncbi.eFetch.Sequences.MessageEFetchRequest preq = new EhuBio.Database.Ncbi.eFetch.Sequences.MessageEFetchRequest();
EhuBio.Database.Ncbi.eFetch.Sequences.MessageEFetchResult pres;
//int count = 1;
while ((line = rd.ReadLine()) != null)
{
fields = line.Split(sep1);
if (!fields[2].Contains("/") || fields[3].Length == 0 || fields[3] != fields[4])
{
continue;
}
v = new Variant();
if (ids.Contains(fields[0]))
{
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);
}
}
