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 List <WregexResult> Search(string seq, string id, ResultType type)
{
List <WregexResult> results = new List <WregexResult>();
Match m = mRegex.Match(seq);
if (!m.Success)
{
return(results);
}
WregexResult result;
do
{
result = new WregexResult();
//result.Id = id + "@" + m.Index;
result.Entry = id;
result.Index = m.Index;
result.Combinations = 1;
result.Groups = new List <string>();
result.Match = m.Value;
result.Length = m.Length;
if (mMaxLength == 0 || m.Length > mMaxLength)
{
mMaxLength = m.Length;
}
if (mMinLength == 0 || m.Length < mMinLength)
{
mMinLength = m.Length;
}
for (int i = 1; i < m.Groups.Count; i++)
{
result.Groups.Add(m.Groups[i].Value);
}
result.Score = mPssm != null?mPssm.GetScore(result) : 0.0;
result.Type = type;
results.Add(result);
m = mRegex.Match(seq, result.Index + 1);
} while(m.Success);
if (mGrouping)
{
return(Filter(results));
}
return(results);
}
private string GetID(List <Variant> m, WregexResult r, bool original)
{
string id = "";
foreach (Variant v in m)
{
if ((int)v.pos >= r.Index && (int)v.pos < (r.Index + r.Length))
{
if (original || r.Match[(int)v.pos - r.Index] == v.mut)
{
id += "-" + v.orig + (v.pos + 1).ToString() + v.mut;
}
}
}
return(id);
}
private List <List <Variant> > GetMutations(Fasta seq, WregexResult r)
{
int i1 = 0, i2;
while (i1 < seq.mVariants.Count && (int)seq.mVariants[i1].pos < r.Index)
{
i1++;
}
i2 = i1;
while (i2 < seq.mVariants.Count && (int)seq.mVariants[i2].pos < (r.Index + r.Length))
{
i2++;
}
return(GetMutations(seq, i1, i2 - 1));
}
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;
}
private List <WregexResult> GetTotalVariantsResults(Fasta seq)
{
List <WregexResult> ret = new List <WregexResult>();
WregexResult[] orig, mut;
bool found;
int i, j;
List <List <Variant> > mutations;
string id;
// Original (without mutations)
orig = mRegex.Search(seq.mSequence, seq.ID + "-orig").ToArray();
ret.AddRange(orig);
// Lost
foreach (WregexResult r in orig)
{
/*if( r.Entry.Contains("NP_002968.1") )
* Console.WriteLine( "KK" );*/
mutations = GetMutations(seq, r);
foreach (List <Variant> m in mutations)
{
if (GetVariantsResults(seq.ID, r.Match, r.Index, m).Count == 0)
{
WregexResult r2 = r;
r2.Type = ResultType.Lost;
r2.Entry = r2.Entry.Replace("-orig", "-lost") + GetID(m, r, true);
ret.Add(r2);
}
}
}
// Mutations
mutations = GetMutations(seq, mRegex.MaxLength);
mut = GetVariantsResults(seq.ID, seq.mSequence, mutations).ToArray();
for (i = 0; i < mut.Length; i++)
{
// Filter duplicates
found = false;
foreach (WregexResult r in ret)
{
if (r.Index == mut[i].Index && r.Match.Equals(mut[i].Match))
{
found = true;
break;
}
}
if (found)
{
continue;
}
// Assign names
id = GetID(seq.mVariants, mut[i]);
if (id.Length == 0)
{
continue;
}
mut[i].Entry += id;
// Gained
found = false;
for (j = 0; j < orig.Length; j++)
{
if (mut[i].Index == orig[j].Index)
{
found = true;
break;
}
}
if (!found)
{
mut[i].Type = ResultType.Gained;
}
ret.Add(mut[i]);
}
return(ret.Count == 0 ? null : ret);
}
private string GetID(List <Variant> m, WregexResult r)
{
return(GetID(m, r, false));
}
public List<WregexResult> Search( string seq, string id, ResultType type )
{
List<WregexResult> results = new List<WregexResult>();
Match m = mRegex.Match(seq);
if( !m.Success )
return results;
WregexResult result;
do {
result = new WregexResult();
//result.Id = id + "@" + m.Index;
result.Entry = id;
result.Index = m.Index;
result.Combinations = 1;
result.Groups = new List<string>();
result.Match = m.Value;
result.Length = m.Length;
if( mMaxLength == 0 || m.Length > mMaxLength )
mMaxLength = m.Length;
if( mMinLength == 0 || m.Length < mMinLength )
mMinLength = m.Length;
for( int i = 1; i < m.Groups.Count; i++ )
result.Groups.Add( m.Groups[i].Value );
result.Score = mPssm != null ? mPssm.GetScore(result) : 0.0;
result.Type = type;
results.Add( result );
m = mRegex.Match( seq, result.Index + 1 );
} while( m.Success );
if( mGrouping )
return Filter(results);
return results;
}
private List<List<Variant>> GetMutations( Fasta seq, WregexResult r )
{
int i1 = 0, i2;
while( i1 < seq.mVariants.Count && (int)seq.mVariants[i1].pos < r.Index )
i1++;
i2 = i1;
while( i2 < seq.mVariants.Count && (int)seq.mVariants[i2].pos < (r.Index+r.Length) )
i2++;
return GetMutations( seq, i1, i2-1 );
}
private string GetID( List<Variant> m, WregexResult r, bool original )
{
string id = "";
foreach( Variant v in m )
if( (int)v.pos >= r.Index && (int)v.pos < (r.Index+r.Length) )
if( original || r.Match[(int)v.pos-r.Index] == v.mut )
id += "-" + v.orig + (v.pos+1).ToString() + v.mut;
return id;
}
private string GetID( List<Variant> m, WregexResult r )
{
return GetID( m, r, false );
}