} // SrSlSlSr
public static Dictionary <DeltaKey, int> CreateScanVerticalDeltaTable()
{
Dictionary <DeltaKey, int> deltaTable = CreateInitialTable(-1);
// Want to match lefts and rights mostly
DeltaKey key = new DeltaKey("Su", "Su");
deltaTable[key] = 1;
key = new DeltaKey("Sd", "Sd");
deltaTable[key] = 1;
foreach (Token y in tokens)
{
// Penalise any kind of insertions
key = new DeltaKey("", Wordbook.TokenToString(y));
deltaTable[key] = -10;
// and deletions
key = new DeltaKey(Wordbook.TokenToString(y), "");
deltaTable[key] = -10;
}
return(deltaTable);
} // SuSdSdSu
} // SuSdSdSu
public static Dictionary <DeltaKey, int> CreateScanCornerDeltaTable()
{
Dictionary <DeltaKey, int> deltaTable = CreateInitialTable(-5);
// Want to match lefts and rights, ups and down
DeltaKey key = new DeltaKey("Su", "Su");
deltaTable[key] = 1;
key = new DeltaKey("Sd", "Sd");
deltaTable[key] = 1;
key = new DeltaKey("Sl", "Sl");
deltaTable[key] = 1;
key = new DeltaKey("Sr", "Sr");
deltaTable[key] = 1;
foreach (Token y in tokens)
{
// Penalise any kind of insertions
key = new DeltaKey("", Wordbook.TokenToString(y));
deltaTable[key] = -10;
// and deletions
key = new DeltaKey(Wordbook.TokenToString(y), "");
deltaTable[key] = -10;
}
return(deltaTable);
} // SrSlSuSd, and all variants
public static Dictionary <DeltaKey, int> CreateCompareVerticalDeltaTable() // SuSdSuSd
{
Dictionary <DeltaKey, int> deltaTable = CreateInitialTable(-2);
// Want to match lefts and rights mostly, perfect match == 8
DeltaKey key = new DeltaKey("Su", "Su");
deltaTable[key] = 2;
key = new DeltaKey("Sd", "Sd");
deltaTable[key] = 2;
foreach (Token y in tokens)
{
// Penalise any kind of insertions
key = new DeltaKey("", Wordbook.TokenToString(y));
deltaTable[key] = -2;
// Penalise any kind of deletion
key = new DeltaKey(Wordbook.TokenToString(y), "");
deltaTable[key] = -2;
}
// Insertions of short lefts and rights are okay
key = new DeltaKey("", "Sl");
deltaTable[key] = -1;
key = new DeltaKey("", "Sr");
deltaTable[key] = -1;
// Therefore, for one insertion of an up or down, the threshold will be 7 or higher.
// Perfect match is 8, minus 1 for 1 insertion of an Su or Sd.
return(deltaTable);
}
} // SrSrSrSrLl
public static Dictionary <DeltaKey, int> CreateLongLineDeltaTable()
{
Dictionary <DeltaKey, int> deltaTable = CreateInitialTable(-1);
DeltaKey key = new DeltaKey("Sr", "Sr");
deltaTable[key] = 5;
key = new DeltaKey("Sr", "Mr");
deltaTable[key] = 2;
key = new DeltaKey("Sr", "Sl");
deltaTable[key] = 0;
key = new DeltaKey("Ll", "Ll");
deltaTable[key] = 10;
key = new DeltaKey("Ll", "");
deltaTable[key] = -10;
foreach (Token y in tokens)
{
key = new DeltaKey("", Wordbook.TokenToString(y));
deltaTable[key] = -10;
}
return(deltaTable);
} // SrSrSrSrLl
// Returns a delta table all initialised to -1.
public static Dictionary <DeltaKey, int> CreateInitialTable(int initialValue)
{
Dictionary <DeltaKey, int> deltaTable = new Dictionary <DeltaKey, int>();
// Add all possibilities
foreach (Token x in tokens)
{
foreach (Token y in tokens)
{
DeltaKey key = new DeltaKey(Wordbook.TokenToString(x), Wordbook.TokenToString(y));
deltaTable[key] = initialValue;
}
}
// Add epsilons for deletion and insertions
foreach (Token y in tokens)
{
DeltaKey key = new DeltaKey("", Wordbook.TokenToString(y));
deltaTable[key] = -1;
}
foreach (Token x in tokens)
{
DeltaKey key = new DeltaKey(Wordbook.TokenToString(x), "");
deltaTable[key] = -1;
}
return(deltaTable);
}
// Local Alignment for ATOMS via Smith–Waterman algorithm
// Returns a list tuples that show the position (startIndex, length) of every match of x in the longer string y, (that is above the threshold).
static public List <Tuple <int, int> > GetLocationsOfLocalMatches(List <Token> x, List <Token> y, Dictionary <DeltaKey, int> delta, int threshold)
{
// Create table
//#############
int numColumns = x.Count + 1;
int numRows = y.Count + 1;
int[,] table = new int[numColumns, numRows];
// Zero out first row and column.
for (int column = 0; column < numColumns; column++)
{
table[column, 0] = 0;
}
for (int row = 0; row < numRows; row++)
{
table[0, row] = 0;
}
// Starting at (1, 1) going left to right, to bottom fill in the spaces in the table.
for (int row = 1; row < numRows; row++)
{
for (int column = 1; column < numColumns; column++)
{
int left = table[column - 1, row] + delta[new DeltaKey(Wordbook.TokenToString(x[column - 1]), "")]; // deletion
int top = table[column, row - 1] + delta[new DeltaKey("", Wordbook.TokenToString(y[row - 1]))];; // insertion
int topLeft = table[column - 1, row - 1] + delta[new DeltaKey(Wordbook.TokenToString(x[column - 1]), Wordbook.TokenToString(y[row - 1]))]; // match or replace
table[column, row] = Max(left, top, topLeft, 0);
}
}
//PrintTable(table, numColumns, numRows);
Tuple <int, int> maxLocation = FindMaxInTable(table, numColumns, numRows);
// Display all substring matches that are above the threshold
var startingLocations = FindMatchLocationsAboveThreshold(table, numColumns, numRows, threshold);
List <Tuple <int, int> > subStringLocations = new List <Tuple <int, int> >();
foreach (var location in startingLocations)
{
Tuple <int, int> subStringPosition = FindMatchingSubstring(table, location, x, y, delta);
int subStringLength = subStringPosition.Item2 - subStringPosition.Item1;
Tuple <int, int> subStringLocation = new Tuple <int, int>(subStringPosition.Item1, subStringLength);
subStringLocations.Add(subStringLocation);
}
return(subStringLocations);
}
} // SuSuSuSu
public static Dictionary <DeltaKey, int> CreateStringRightDeltaTable()
{
Dictionary <DeltaKey, int> deltaTable = CreateInitialTable(-1);
// Want to match only rights. Perfect match would score 8.
DeltaKey key = new DeltaKey("Sr", "Sr");
deltaTable[key] = 2;
// Replacing Sr with any number of Mr is okay too.
key = new DeltaKey("Sr", "Mr");
deltaTable[key] = 2;
// Replacing a short right in the string with a short left is okay.
key = new DeltaKey("Sr", "Sl");
deltaTable[key] = 1;
// Short ups and downs are okay
key = new DeltaKey("Sr", "Su");
deltaTable[key] = 1;
key = new DeltaKey("Sr", "Sd");
deltaTable[key] = 1;
// If there is one replacement of an Sr either with Sl Su Sd, and the rest match with Sr, then the matching score would be 7. 6 for two replacements, etc.
// We want to allow only one replacement somewhere in the string. So we will want a threshold of 7 when finding matches.
// todo: need a way to specify the first and last tokens in the atom cannot change
// perhaps something like Sr - - Sr, where Sr matching with Sr score is high and whatever the dummy - char is has an equally high score for being replace with Sr, whilst slightly lower score for being replaced with Su, Sl, Sd.
foreach (Token y in tokens)
{
// Penalise any kind of insertions
key = new DeltaKey("", Wordbook.TokenToString(y));
deltaTable[key] = -10;
// Penalise any kind of deletion
key = new DeltaKey(Wordbook.TokenToString(y), "");
deltaTable[key] = -10;
}
return(deltaTable);
} // SrSrSrSr, see notes for matching information
static private Tuple <int, int> FindMatchingSubstring(int[,] table, Tuple <int, int> maxLocation, List <Token> x, List <Token> y, Dictionary <DeltaKey, int> delta)
{
int column = maxLocation.Item1;
int row = maxLocation.Item2;
bool searching = true;
while (searching)
{
// Keep searching until we git a 0 in the table.
if (table[column, row] == 0)
{
searching = false;
break;
}
// Otherwise, keep moving up the table following the path we took to get here.
int topLeft = table[column - 1, row - 1] + delta[new DeltaKey(Wordbook.TokenToString(x[column - 1]), Wordbook.TokenToString(y[row - 1]))]; // match or replace
if (table[column, row] == topLeft)
{
column--;
row--;
continue;
}
int left = table[column - 1, row] + delta[new DeltaKey(Wordbook.TokenToString(x[column - 1]), "")]; // deletion
int top = table[column, row - 1] + delta[new DeltaKey("", Wordbook.TokenToString(y[row - 1]))]; // insertion
if (table[column, row] == left)
{
column--;
continue;
}
if (table[column, row] == top)
{
row--;
continue;
}
}
return(Tuple.Create(row, maxLocation.Item2));
}
} // SdSdSdSd
public static Dictionary <DeltaKey, int> CreateStringLeftDeltaTable()
{
Dictionary <DeltaKey, int> deltaTable = CreateInitialTable(-1);
// Want to match only lefts. Perfect match would score 8.
DeltaKey key = new DeltaKey("Sl", "Sl");
deltaTable[key] = 2;
// Replacing Sl with any number of Ml is okay too.
key = new DeltaKey("Sl", "Ml");
deltaTable[key] = 2;
// Short left rights and ups are okay.
key = new DeltaKey("Sl", "Sd");
deltaTable[key] = 1;
key = new DeltaKey("Sl", "Sr");
deltaTable[key] = 1;
key = new DeltaKey("Sl", "Su");
deltaTable[key] = 1;
foreach (Token y in tokens)
{
// Penalise any kind of insertions
key = new DeltaKey("", Wordbook.TokenToString(y));
deltaTable[key] = -10;
// Penalise any kind of deletion
key = new DeltaKey(Wordbook.TokenToString(y), "");
deltaTable[key] = -10;
}
return(deltaTable);
} // SlSlSlSl