//http://rosalind.info/problems/pdst/
public CreatingADistanceMatrix()
{
string[] dnaStrings = FASTAToDictionary.Convert(File.ReadAllLines(@"C:\code\dataset.txt").ToList()).Values.ToArray();
double stringLength = dnaStrings[0].Length;
for (int i = 0; i < dnaStrings.Length; i++)
{
for (int j = 0; j < dnaStrings.Length; j++)
{
double pDistance = 0;
if (i != j)
{
for (int k = 0; k < stringLength; k++)
{
if (dnaStrings[i][k] != dnaStrings[j][k])
{
pDistance++;
}
}
}
Console.Write((pDistance / stringLength).ToString("0.00000") + " ");
}
Console.Write(Environment.NewLine);
}
}
//http://rosalind.info/problems/lcsm/
public FindingASharedMotif()
{
List <string> input = File.ReadAllLines(@"C:\code\dataset.txt").ToList();
var dnaStrings = FASTAToDictionary.Convert(input).Values.ToList();
dnaStrings = dnaStrings.OrderBy(s => s.Length).ToList();
List <string> subStrings = new List <string>();
for (int i = 0; i < dnaStrings[0].Length; i++)
{
for (int j = 1; j <= dnaStrings[0].Length - i; j++)
{
subStrings.Add(dnaStrings[0].Substring(i, j));
}
}
subStrings = subStrings.OrderByDescending(s => s.Length).ToList();
foreach (string subString in subStrings)
{
if (dnaStrings.All(s => s.Contains(subString)))
{
Console.WriteLine(subString);
return;
}
}
}
//http://rosalind.info/problems/edit/
public EditDistance()
{
List <string> dnaStrings = FASTAToDictionary.Convert(File.ReadAllLines(@"C:\code\dataset.txt").ToList()).Values.ToList();
int changeCount = Distance(dnaStrings[0], dnaStrings[1], dnaStrings[0].Length - 1, dnaStrings[1].Length - 1, new Dictionary <KeyValuePair <int, int>, int>());
Console.WriteLine(changeCount);
}
//http://rosalind.info/problems/splc/
public RnaSplicing()
{
List <string> dnaStrings = FASTAToDictionary.Convert(File.ReadAllLines(@"C:\code\dataset.txt").ToList()).Values.ToList();
string dnaString = dnaStrings[0];
for (int i = 1; i < dnaStrings.Count; i++)
{
dnaString = dnaString.Replace(dnaStrings[i], "");
}
dnaString = dnaString.Replace('T', 'U');
using (var stringReader = new StringReader(dnaString))
{
string aa = "";
while (aa != "Stop")
{
char[] buffer = new char[3];
stringReader.ReadBlock(buffer, 0, 3);
aa = string.Join("", buffer).ConvertCodon();
if (aa != "Stop")
{
Console.Write(aa);
}
}
}
}
//http://rosalind.info/problems/lcsq/
public FindingASharedSplicedMotifAlternative2()
{
List <string> dnaStrings = FASTAToDictionary.Convert(File.ReadAllLines(@"C:\code\dataset.txt").ToList()).Values.ToList();
blu(dnaStrings[0], dnaStrings[1], "ACGT");
blu(dnaStrings[0], dnaStrings[1], "ACTG");
blu(dnaStrings[0], dnaStrings[1], "AGCT");
blu(dnaStrings[0], dnaStrings[1], "AGTC");
blu(dnaStrings[0], dnaStrings[1], "ATGC");
blu(dnaStrings[0], dnaStrings[1], "ATCG");
blu(dnaStrings[0], dnaStrings[1], "CAGT");
blu(dnaStrings[0], dnaStrings[1], "CATG");
blu(dnaStrings[0], dnaStrings[1], "CGAT");
blu(dnaStrings[0], dnaStrings[1], "CGTA");
blu(dnaStrings[0], dnaStrings[1], "CTAG");
blu(dnaStrings[0], dnaStrings[1], "CTGA");
blu(dnaStrings[0], dnaStrings[1], "GCAT");
blu(dnaStrings[0], dnaStrings[1], "GCTA");
blu(dnaStrings[0], dnaStrings[1], "GACT");
blu(dnaStrings[0], dnaStrings[1], "GATC");
blu(dnaStrings[0], dnaStrings[1], "GTCA");
blu(dnaStrings[0], dnaStrings[1], "GTAC");
blu(dnaStrings[0], dnaStrings[1], "TCGA");
blu(dnaStrings[0], dnaStrings[1], "TCAG");
blu(dnaStrings[0], dnaStrings[1], "TAGC");
blu(dnaStrings[0], dnaStrings[1], "TACG");
blu(dnaStrings[0], dnaStrings[1], "TGAC");
blu(dnaStrings[0], dnaStrings[1], "TGCA");
}
//http://rosalind.info/problems/kmp/
public SpeedingUpMotifFinding()
{
string input = FASTAToDictionary.Convert(File.ReadAllLines(@"C:\code\dataset.txt").ToList()).First().Value;
int sequence = 0;
for (int i = 0; i < input.Length; i++)
{
if (i != 0 && input[sequence] == input[i])
{
sequence++;
}
else if (sequence != 0)
{
for (int j = sequence; j >= 0; j--)
{
if (input.Substring(0, j).Equals(input.Substring(i - j + 1, j)))
{
sequence = j;
break;
}
}
}
else
{
sequence = 0;
}
Console.Write(sequence + " ");
}
}
//http://rosalind.info/problems/pmch/
public PerfectMatchingAndRnaSecondaryStructuresAlternative()
{
string input = FASTAToDictionary.Convert(File.ReadAllLines(@"C:\code\dataset.txt").ToList()).First().Value;
_perfectLength = input.Length / 2;
List <List <int> > validConnections = ValidConnections(input);
List <List <int> > perfectMatchings = new List <List <int> >();
//PerfectMatchings(validConnections, new List<int>(), 0);
UInt64 G = (ulong)input.Count(c => c == 'G');
UInt64 A = (ulong)input.Count(c => c == 'A');
ulong d = 121645100408832;
ulong e = 20922789888;
ulong f = e * d;
Console.WriteLine( );
//UInt64 Fg = Factorial(G);
//UInt64 Fa = Factorial(A);
//UInt64 result = Fg * Fa;
//Console.WriteLine(result);
////Console.WriteLine(_count);
}
//http://rosalind.info/problems/kmer/
public KMerComposition()
{
List <string> dnaStrings = FASTAToDictionary.Convert(File.ReadAllLines(@"C:\code\dataset.txt").ToList()).Values.ToList();
string symbols = "";
for (int i = 0; i < dnaStrings[0].Length; i++)
{
if (symbols.All(c => c != dnaStrings[0][i]))
{
symbols += dnaStrings[0][i];
}
}
var symbolList = symbols.Select(c => c.ToString()).ToList();
symbolList.Sort();
var kmers = new Dictionary <string, int>();
Solve(kmers, symbolList, "", 4);
for (int i = 0; i < dnaStrings[0].Length - 3; i++)
{
kmers[dnaStrings[0].Substring(i, 4)]++;
}
Console.Write(string.Join(" ", kmers.Values));
}
//http://rosalind.info/problems/lcsq/
public FindingASharedSplicedMotif()
{
List <string> dnaStrings = FASTAToDictionary.Convert(File.ReadAllLines(@"C:\code\dataset.txt").ToList()).Values.ToList();
dod(dnaStrings[0], dnaStrings[1]);
Console.WriteLine(_subsequence);
Console.WriteLine("done");
}
//http://rosalind.info/problems/long/
public GenomeAssemblyAsShortestSuperstring()
{
List <string> dnaStrings = FASTAToDictionary.Convert(File.ReadAllLines(@"C:\code\dataset.txt").ToList()).Values.ToList();
string superstring = dnaStrings[0];
dnaStrings.RemoveAt(0);
while (dnaStrings.Any())
{
int bestPrefixLength = 0;
int bestSuffixLength = 0;
int bestPrefix = -1;
int bestSuffix = -1;
for (int i = 0; i < dnaStrings.Count; i++)
{
int overlap = dnaStrings[i].Length;
while (overlap > dnaStrings[i].Length / 2 && (overlap > bestPrefixLength || overlap > bestSuffixLength))
{
if (overlap > bestPrefixLength && superstring.StartsWith(dnaStrings[i].Substring(dnaStrings[i].Length - overlap)))
{
bestPrefixLength = overlap;
bestPrefix = i;
}
else if (overlap > bestSuffixLength && superstring.EndsWith(dnaStrings[i].Substring(0, overlap)))
{
bestSuffixLength = overlap;
bestSuffix = i;
}
else
{
overlap--;
}
}
}
if (bestPrefix >= 0)
{
superstring = dnaStrings[bestPrefix] + superstring.Substring(bestPrefixLength);
}
if (bestSuffix >= 0)
{
superstring = superstring + dnaStrings[bestSuffix].Substring(bestSuffixLength);
}
dnaStrings.RemoveAt(Math.Max(bestPrefix, bestSuffix));
if (Math.Min(bestPrefix, bestSuffix) >= 0)
{
dnaStrings.RemoveAt(Math.Min(bestPrefix, bestSuffix));
}
}
Console.WriteLine(superstring);
}
//http://rosalind.info/problems/corr/
public ErrorCorrectionInReads()
{
string[] dnaStrings = FASTAToDictionary.Convert(File.ReadAllLines(@"C:\code\dataset.txt").ToList()).Values.ToArray();
var valid = new List <string>();
var invalid = new List <string>();
foreach (string dnaString in dnaStrings)
{
bool isValid = false;
string reverse = ComplementingAStrandOfDna.ReverseCompliment(dnaString, false);
for (int i = 0; i < valid.Count; i++)
{
if (dnaString.Equals(valid[i]) || reverse.Equals(valid[i]))
{
isValid = true;
break;
}
}
if (isValid)
{
continue;
}
for (int i = 0; i < invalid.Count; i++)
{
if (dnaString.Equals(invalid[i]) || reverse.Equals(invalid[i]))
{
valid.Add(dnaString);
valid.Add(reverse);
invalid.RemoveAt(i);
isValid = true;
break;
}
}
if (isValid == false)
{
invalid.Add(dnaString);
}
}
foreach (string inv in invalid)
{
foreach (var val in valid)
{
if (CountingPointMutations.GetHammingDistance(inv, val) == 1)
{
Console.WriteLine(inv + "->" + val);
}
}
}
}
//http://rosalind.info/problems/pmch/
public PerfectMatchingAndRnaSecondaryStructures()
{
string input = FASTAToDictionary.Convert(File.ReadAllLines(@"C:\code\dataset.txt").ToList()).First().Value;
_perfectLength = input.Length / 2;
var result = new List <List <KeyValuePair <int, int> > >();
Matches(input, result, new List <KeyValuePair <int, int> >());
Console.WriteLine(result.Count);
}
//http://rosalind.info/problems/sseq/
public FindingASplicedMotif()
{
List <string> dnaStrings = FASTAToDictionary.Convert(File.ReadAllLines(@"C:\code\dataset.txt").ToList()).Values.ToList();
int index = 0;
foreach (char c in dnaStrings[1])
{
index = dnaStrings[0].IndexOf(c, index) + 1;
Console.Write(index + " ");
}
}
//http://rosalind.info/problems/ctea/
public CountingOptimalAlignments()
{
List <string> dnaStrings = FASTAToDictionary.Convert(File.ReadAllLines(@"C:\code\dataset.txt").ToList()).Values.ToList();
var pairs = new Dictionary <KeyValuePair <int, int>, KeyValuePair <int, int> >();
_s1 = dnaStrings[0];
_s2 = dnaStrings[1];
int changeCount = EditDistanceAlignment.GenerateAlignmentPathPairs(_s1, _s2, _s1.Length - 1, _s2.Length - 1, pairs);
ulong count = CountPaths(pairs, pairs.Last().Key, new Dictionary <KeyValuePair <int, int>, ulong>());
Console.WriteLine(count);
}
//http://rosalind.info/problems/lcsq/
public FindingASharedSplicedMotifStolen()
{
List <string> dnaStrings = FASTAToDictionary.Convert(File.ReadAllLines(@"C:\code\dataset.txt").ToList()).Values.ToList();
var pairs = new Dictionary <KeyValuePair <int, int>, int>();
dod(dnaStrings[0], dnaStrings[1], dnaStrings[0].Length - 1, dnaStrings[1].Length - 1, pairs);
string sequence = reconstruct(dnaStrings[0], dnaStrings[1], dnaStrings[0].Length - 1, dnaStrings[1].Length - 1, pairs);
sequence = string.Join("", sequence.Reverse());
Console.WriteLine(sequence.Length);
Console.WriteLine(sequence);
//Console.WriteLine(Environment.NewLine);
//Console.WriteLine(bluh(dnaStrings[0], dnaStrings[1], dnaStrings[0].Length - 1, dnaStrings[1].Length - 1));
}
//http://rosalind.info/problems/glob/
public GlobalAlignmentWithScoringMatrix()
{
List <string> dnaStrings = FASTAToDictionary.Convert(File.ReadAllLines(@"C:\code\dataset.txt").ToList()).Values.ToList();
var pairs = new Dictionary <KeyValuePair <int, int>, KeyValuePair <int, int> >();
var s1Chars = new List <char>(dnaStrings[0]);
var s2Chars = new List <char>(dnaStrings[1]);
int changeCount = GenerateAlignmentPathPairs(dnaStrings[0], dnaStrings[1], dnaStrings[0].Length - 1, dnaStrings[1].Length - 1, pairs);
AddHyphens(s1Chars, s2Chars, pairs, pairs.Last().Key);
Console.WriteLine(changeCount);
Console.WriteLine(string.Join("", s1Chars));
Console.WriteLine(string.Join("", s2Chars));
}
//http://rosalind.info/problems/cons/
public ConcensusAndProfile()
{
List <string> dnaStrings = FASTAToDictionary.Convert(File.ReadAllLines(@"C:\code\dataset.txt").ToList()).Values.ToList();
int[] A = new int[dnaStrings[0].Length];
int[] C = new int[dnaStrings[0].Length];
int[] G = new int[dnaStrings[0].Length];
int[] T = new int[dnaStrings[0].Length];
for (int i = 0; i < dnaStrings[0].Length; i++)
{
string concensus = "";
for (int j = 0; j < dnaStrings.Count; j++)
{
concensus += dnaStrings[j][i];
}
A[i] = concensus.Count(c => c == 'A');
C[i] = concensus.Count(c => c == 'C');
G[i] = concensus.Count(c => c == 'G');
T[i] = concensus.Count(c => c == 'T');
char c2 = 'A';
int max = A[i];
if (C[i] > max)
{
c2 = 'C';
max = C[i];
}
if (G[i] > max)
{
c2 = 'G';
max = G[i];
}
if (T[i] > max)
{
c2 = 'T';
}
Console.Write(c2);
}
Console.Write(Environment.NewLine);
WriteCharacterArray(A, 'A');
WriteCharacterArray(C, 'C');
WriteCharacterArray(G, 'G');
WriteCharacterArray(T, 'T');
}
//http://rosalind.info/problems/orf/
public OpenReadingFrames()
{
var proteins = new List <ProteinString>();
string input = FASTAToDictionary.Convert(File.ReadAllLines(@"C:\code\dataset.txt").ToList()).First().Value.Replace('T', 'U');
FindProteinOpenFrames(proteins, input);
FindProteinOpenFrames(proteins, input.Substring(1));
FindProteinOpenFrames(proteins, input.Substring(2));
input = ComplementingAStrandOfDna.ReverseCompliment(input, true);
FindProteinOpenFrames(proteins, input);
FindProteinOpenFrames(proteins, input.Substring(1));
FindProteinOpenFrames(proteins, input.Substring(2));
proteins.Select(p => p.Protein).Distinct().ToList().ForEach(p => Console.WriteLine(p));
}
//http://rosalind.info/problems/cat/
public CatalanNumbersAndRnaSecondaryStructures()
{
string input = FASTAToDictionary.Convert(File.ReadAllLines(@"C:\code\dataset.txt").ToList()).First().Value;
List <char> nodes = input.ToCharArray().ToList();
var store = new Dictionary <KeyValuePair <int, int>, long>();
long result = 0;
for (int i = 1; i < nodes.Count; i += 2)
{
var pair = new KeyValuePair <int, int>(0, i);
long rtn = CatalanNumbers(store, nodes, pair, i);
result = (result + rtn) % 1000000;
}
Console.WriteLine(result);
}
//http://rosalind.info/problems/grph/
public OverlapGraphs()
{
Dictionary <string, string> dnaStrings = FASTAToDictionary.Convert(File.ReadAllLines(@"C:\code\dataset.txt").ToList());
foreach (KeyValuePair <string, string> dnaString in dnaStrings)
{
foreach (KeyValuePair <string, string> dnaStringMatch in dnaStrings)
{
if (dnaString.Value.Equals(dnaStringMatch.Value) == false)
{
if (dnaString.Value.Substring(dnaString.Value.Length - 3, 3).Equals(dnaStringMatch.Value.Substring(0, 3)))
{
Console.WriteLine(dnaString.Key + " " + dnaStringMatch.Key);
}
}
}
}
}
//http://rosalind.info/problems/lcsq/
public FindingASharedSplicedMotifAlternative()
{
List <string> dnaStrings = FASTAToDictionary.Convert(File.ReadAllLines(@"C:\code\dataset.txt").ToList()).Values.ToList();
for (int i = 0; i < dnaStrings[0].Length; i++)
{
//for (int j = 0; j < 500; j++)
{
if (dnaStrings[0].Length - i > _result.Length)
{
dod(new List <char>(dnaStrings[0]), new List <char>(dnaStrings[1]), i, 1);
dod(new List <char>(dnaStrings[0]), new List <char>(dnaStrings[1]), i, 2);
dod(new List <char>(dnaStrings[0]), new List <char>(dnaStrings[1]), i, 3);
}
}
}
Console.WriteLine(_index + "(" + _result.Length + ")" + " done: ");
Console.WriteLine(_result);
}
//http://rosalind.info/problems/motz/
public MotzkinNumbersAndRnaSecondaryStructures()
{
//Add all the most obvious ones as far as you can one at a time, then step back and retread all the way through
string input = FASTAToDictionary.Convert(File.ReadAllLines(@"C:\code\dataset.txt").ToList()).First().Value;
List <char> nodes = input.ToCharArray().ToList();
var store = new Dictionary <KeyValuePair <int, int>, long>();
long result = 0;
for (int i = 1; i < nodes.Count; i += 2)
{
var pair = new KeyValuePair <int, int>(0, i);
long rtn = MotzkinNumbers(store, nodes, pair, i);
result = (result + rtn) % 1000000;
}
Console.WriteLine(1 + result); //add one for no connections
}
//http://rosalind.info/problems/revp/
public LocatingRestrictionSites()
{
string input = FASTAToDictionary.Convert(File.ReadAllLines(@"C:\code\dataset.txt").ToList()).First().Value;
string reversed = ComplementingAStrandOfDna.ReverseCompliment(input, false);
for (int i = 0; i < input.Length - 3; i++)
{
for (int j = 4; j <= 12; j++)
{
if (i + j > input.Length)
{
break;
}
if (input.Substring(i, j).Equals(reversed.Substring(input.Length - (i + j), j)))
{
Console.WriteLine(i + 1 + " " + j);
}
}
}
}
//http://rosalind.info/problems/mmch/
public MaximumMatchingsAndRnaSecondaryStructures()
{
//.NET still can't handle big numbers~
string input = FASTAToDictionary.Convert(File.ReadAllLines(@"C:\code\dataset.txt").ToList()).First().Value;
UInt64 G = (ulong)input.Count(c => c == 'G');
UInt64 C = (ulong)input.Count(c => c == 'C');
UInt64 A = (ulong)input.Count(c => c == 'A');
UInt64 U = (ulong)input.Count(c => c == 'U');
UInt64 maxGC = Math.Max(G, C);
UInt64 minGC = Math.Min(G, C);
UInt64 maxAU = Math.Max(A, U);
UInt64 minAU = Math.Min(A, U);
UInt64 Fg = Factorial(maxGC, maxGC - minGC);
UInt64 Fa = Factorial(maxAU, maxAU - minAU);
UInt64 result = Fg * Fa;
Console.WriteLine(result);
}
//http://rosalind.info/problems/tran/
public TransitionsAndTransversions()
{
string [] dnaStrings = FASTAToDictionary.Convert(File.ReadAllLines(@"C:\code\dataset.txt").ToList()).Values.ToArray();
double transitions = 0;
double transversions = 0;
for (int i = 0; i < dnaStrings[0].Length; i++)
{
char a = dnaStrings[0][i];
char b = dnaStrings[1][i];
if (_transitions.ContainsKey(a) && _transitions[a] == b)
{
transitions++;
}
else if (a != b)
{
transversions++;
}
}
Console.WriteLine(transitions / transversions);
}
