public void OverlapGraphTest()
{
string input = @">Rosalind_0498
AAATAAA
>Rosalind_2391
AAATTTT
>Rosalind_2323
TTTTCCC
>Rosalind_0442
AAATCCC
>Rosalind_5013
GGGTGGG";
var expected = new List <Tuple <string, string> > ()
{
Tuple.Create("Rosalind_0498", "Rosalind_2391"),
Tuple.Create("Rosalind_0498", "Rosalind_0442"),
Tuple.Create("Rosalind_2391", "Rosalind_2323")
};
var result = new List <Tuple <string, string> > (
OverlapGrapher.calculateGraph(3, FASTA.read(new StringReader(input))));
Assert.AreEqual(expected.Count, result.Count);
foreach (var pair in expected.Zip(result, (a, b) => Tuple.Create(a, b)))
{
Assert.AreEqual(pair.Item1, pair.Item2);
}
}
public void TestConsensus()
{
string input = @">Rosalind_1
ATCCAGCT
>Rosalind_2
GGGCAACT
>Rosalind_3
ATGGATCT
>Rosalind_4
AAGCAACC
>Rosalind_5
TTGGAACT
>Rosalind_6
ATGCCATT
>Rosalind_7
ATGGCACT";
var profile = new Profile(8);
foreach (var pair in FASTA.read(new StringReader(input)))
{
var seq = pair.Item2.Select(b => (Base)Enum.Parse(typeof(Base), b.ToString()));
profile.add(seq);
}
var expectedConcensus = "ATGCAACT".Select(b => (Base)Enum.Parse(typeof(Base), b.ToString()));
IDictionary <Base, IList <uint> > expectedProfile = new Dictionary <Base, IList <uint> > ()
{
{ Base.A, new List <uint> ()
{
5, 1, 0, 0, 5, 5, 0, 0
} },
{ Base.C, new List <uint> ()
{
0, 0, 1, 4, 2, 0, 6, 1
} },
{ Base.G, new List <uint> ()
{
1, 1, 6, 3, 0, 1, 0, 0
} },
{ Base.T, new List <uint> ()
{
1, 5, 0, 0, 0, 1, 1, 6
} }
};
Assert.That(profile.consensus, Is.EqualTo(expectedConcensus));
foreach (Base b in Enum.GetValues(typeof(Base)))
{
Assert.That(
new List <uint>(profile[b]),
Is.EqualTo(expectedProfile[b]));
}
}
public static List <FASTA> DecodeFasta(string data)
{
List <FASTA> fastaList = new List <FASTA>();
string[] splitData = data.Split('>');
foreach (string item in splitData)
{
FASTA fasta = new FASTA();
string[] temp = item.Split('\n');
fasta.ID = temp[0];
}
return(fastaList);
}
private static void SharedMotifCLI(IDictionary <string, ValueObject> args)
{
string filename = args ["<filename>"].ToString();
using (var infile = new FileStream(filename, FileMode.Open))
using (var reader = new StreamReader(infile)) {
var tracker = FASTA.read(reader)
.Select(p => p.Item2)
.Aggregate(new MotifTracker(), (trk, s) => trk.add(s));
Console.WriteLine(
tracker.motifs.Where(x => x.Length == tracker.maxLen).First());
}
}
private static void OverlapGraph(IDictionary <string, ValueObject> args)
{
string filename = args ["<filename>"].ToString();
using (var infile = new FileStream(filename, FileMode.Open))
using (var reader = new StreamReader(infile)) {
var result = OverlapGrapher.calculateGraph(3, FASTA.read(reader));
foreach (var pair in result)
{
Console.WriteLine("{0} {1}", pair.Item1, pair.Item2);
}
}
}
public void TestSharedMotif()
{
const string input = @">Rosalind_1
GATTACA
>Rosalind_2
TAGACCA
>Rosalind_3
ATACA";
const string expected = "AC";
var tracker = FASTA.read(new StringReader(input))
.Aggregate(new MotifTracker(),
(trk, p) => trk.add(p.Item2));
Assert.IsTrue(tracker.motifs.Contains(expected));
}
private static void Consensus(IDictionary <string, ValueObject> args)
{
string filename = args ["<filename>"].ToString();
int size = 0;
using (var infile = new FileStream(filename, FileMode.Open)) {
var reader = new StreamReader(infile);
foreach (var pair in FASTA.read(reader))
{
size = pair.Item2.Length;
}
}
var profile = new Profile((uint)size);
using (var infile = new FileStream(filename, FileMode.Open)) {
var reader = new StreamReader(infile);
foreach (var pair in FASTA.read(reader))
{
var seq = pair.Item2.Select(b => (Base)Enum.Parse(typeof(Base), b.ToString()));
profile.add(seq);
}
}
Console.WriteLine(string.Join("", profile.consensus));
Console.Write("A: ");
Console.WriteLine(string.Join(" ", profile [Base.A]));
Console.Write("C: ");
Console.WriteLine(string.Join(" ", profile [Base.C]));
Console.Write("G: ");
Console.WriteLine(string.Join(" ", profile [Base.G]));
Console.Write("T: ");
Console.WriteLine(string.Join(" ", profile [Base.T]));
}
static void Main(string[] args)
{
string strPath = Environment.GetFolderPath(Environment.SpecialFolder.DesktopDirectory);
string[] files = Directory.GetFiles(strPath + @"\Rosalind", "*.txt"); //Assumes only one txt file in this directory - the data set
string[] lines = { "" };
foreach (string file in files)
{
lines = File.ReadAllLines(file);
}
string data = "";
foreach (var line in lines)
{
data += line;
}
DecodeFasta(data);
FASTA dna = new FASTA();
List <FASTA> fastaList = new List <FASTA>();
foreach (string str in lines)
{
if (str.StartsWith(">"))
{
fastaList.Add(dna);
dna = new FASTA();
dna.ID = str;
}
else
{
dna.DNA += str;
}
}
fastaList.RemoveAt(0); //Remove initial empty FASTA format DNA string
fastaList.Add(dna); //Add last remaining.
FASTA largestGCContent = new FASTA
{
GCContent = 0
};
foreach (var fasta in fastaList)
{
fasta.GCContent = CalculateGCContent(fasta.DNA);
Console.WriteLine(fasta.ID);
Console.WriteLine(fasta.GCContent * 100);
if (fasta.GCContent > largestGCContent.GCContent)
{
largestGCContent.GCContent = fasta.GCContent;
largestGCContent.ID = fasta.ID;
}
}
Console.WriteLine(largestGCContent.ID.Remove(0, 1)); //Remove '>' Character
Console.WriteLine(largestGCContent.GCContent * 100);
Console.ReadLine();
}
