private void Create(string RegexFile, string PssmFile, string DatabaseFile, string VariantsFile, string OutputDir, bool grouping)
{
string line;
// regex
UnixCfg rd = new UnixCfg(RegexFile);
line = rd.ReadUnixLine();
if (line == null)
{
throw new ApplicationException("Empty regex");
}
rd.Close();
if (PssmFile.Length > 0)
{
mRegex = new WregexManager(line, new PSSM(PssmFile), grouping);
}
else
{
mRegex = new WregexManager(line, grouping);
}
// Fasta
mSeqs = new List <Fasta>();
if (DatabaseFile.Contains(".fasta"))
{
LoadFasta(DatabaseFile);
}
else
{
SortedList <string, List <Variant> > list = null;
if (VariantsFile.Length != 0)
{
list = LoadVariants(VariantsFile);
}
LoadXml(DatabaseFile, list);
}
// Sort Variants
foreach (Fasta seq in mSeqs)
{
seq.mVariants.Sort();
}
mDataId = Path.GetFileNameWithoutExtension(DatabaseFile);
mOutputDir = OutputDir;
Directory.CreateDirectory(OutputDir);
}
private void LoadFasta(string path)
{
string line;
//char[] sep = new char[]{'|',' ','\t'};
char[] sep = new char[] { '|' };
Variant v;
UnixCfg rd = new UnixCfg(path);
line = rd.ReadUnixLine();
if (line == null || line[0] != '>')
{
throw new ApplicationException("FASTA header not found");
}
Fasta f = new Fasta(Fasta.Type.Protein, line.Split(sep)[0].Substring(1), "");
do
{
line = rd.ReadUnixLine();
if (line == null || line[0] == '>') // EOF or next element
{
if (f.mSequence.Length == 0)
{
throw new ApplicationException("FASTA sequence not found");
}
mSeqs.Add(f);
if (line != null)
{
f = new Fasta(Fasta.Type.Protein, line.Split(sep)[0].Substring(1), "");
}
}
else if (line.StartsWith("NP_")) // Variant
{
v = new Variant(line);
if (!f.mVariants.Contains(v))
{
f.mVariants.Add(v);
}
}
else // Sequence
{
f.mSequence += line;
}
} while(line != null);
rd.Close();
}
/*public void Load( string filename ) {
* UnixCfg rd = new UnixCfg( filename );
* string line = rd.ReadUnixLine();
* char[] sep = new char[]{ ':' };
* string[] fields;
*
* while( line != null ) {
* PSSMposition pos;
* pos.order = uint.Parse( line.Replace("{","") );
* pos.entries = new List<PSSMentry>();
* line = rd.ReadUnixLine();
* while( line != null && !line.Contains("}") ) {
* PSSMentry entry;
* fields = line.Split( sep );
* entry.condition = fields[0].Trim();
* entry.score = double.Parse( fields[1] );
* pos.entries.Add( entry );
* line = rd.ReadUnixLine();
* }
* if( line != null )
* line = rd.ReadUnixLine();
* m_pssm.Add( pos );
* }
* rd.Close();
* }*/
public void Load(string filename)
{
UnixCfg rd = new UnixCfg(filename);
char[] sep = new char[] { ' ', '\t' };
string[] fields;
PSSMentry entry;
string line = rd.ReadUnixLine();
if (line == null)
{
rd.Close();
return;
}
fields = line.Split(sep);
PSSMposition[] pos = new PSSMposition[fields.Length - 1];
for (uint i = 0; i < pos.Length; i++)
{
pos[i].order = i;
pos[i].entries = new List <PSSMentry>();
}
do
{
fields = line.Split(sep);
entry.condition = fields[0].Trim();
for (uint i = 0; i < pos.Length; i++)
{
entry.score = double.Parse(fields[i + 1]);
pos[i].entries.Add(entry);
}
line = rd.ReadUnixLine();
} while(line != null);
rd.Close();
m_pssm.AddRange(pos);
}