public static void Main(string[] args)
{
//string fileName = "/home/alvydas/Oligonucleotides/Sequencies/Bacteria/Escherichia.coli.gb";
//string fileName = "/home/alvydas/Oligonucleotides/Sequencies/Bacteria/Escherichia.coli.test.gb";
string fileName = "/home/alvydas/Oligonucleotides/Sequencies/Bacteria/Escherichia.coli.short.gb";
//string fileName = "/home/alvydas/Oligonucleotides/Sequencies/Bacteria/Escherichia.coli.short1.gb";
//string fileName = "/home/alvydas/Oligonucleotides/Sequencies/Bacteria/Escherichia.coli.short2.gb";
//string fileName = "/home/alvydas/Oligonucleotides/Sequencies/Bacteria/Escherichia.coli.shuffle.test.gb";
Console.WriteLine($"Calc seq cds trinucs for { Path.GetFileName(fileName) }");
GBSequence gbSequence = new GBSequence(fileName);
var gbSeq = gbSequence.GbSeq();
GBSeqFeatures gBSeqFeatures = new GBSeqFeatures(fileName);
List <GBFeat> cdsFeatures = gBSeqFeatures.FeaturesSeparation();
List <GBFeat> completeFeatures = gBSeqFeatures.CompleteSeparation(cdsFeatures);
List <TrinucDiff> triDiff = new List <TrinucDiff>();
foreach (var item in completeFeatures)
{
string subSeq = string.Empty;
Console.Write($"{item.SeqType}\t{item.SeqStart}\t{item.SeqEnd}\t");
if (item.SeqEnd - item.SeqStart > 40)
{
subSeq = gbSeq.Seq.Substring(item.SeqStart, item.SeqEnd - item.SeqStart + 1);
if (item.SeqType == "CDS" || item.SeqType == "JCDS" || item.SeqType == "NCDS")
{
Console.Write($"{subSeq.Substring(0, 10)}...{subSeq.Substring(subSeq.Length - 10)}\t");
TrinucCalc trinucCalc = new TrinucCalc();
TrinucDiff trinucDiff = trinucCalc.Calculation(item.SeqType, subSeq);
Console.WriteLine($"{ trinucDiff.DiffSum1st2nd.ToString("0.0000") }\t" +
$"{ trinucDiff.DiffSum1st3rd.ToString("0.0000") }\t" +
$"{ trinucDiff.DiffSum2nd3rd.ToString("0.0000") }\t" +
$"{ trinucDiff.DiffSum.ToString("0.0000") }");
if (item.SeqType == "CDS" || item.SeqType == "NCDS")
{
triDiff.Add(new TrinucDiff {
SeqType = item.SeqType,
DiffSum1st2nd = trinucDiff.DiffSum1st2nd,
DiffSum1st3rd = trinucDiff.DiffSum1st3rd,
DiffSum2nd3rd = trinucDiff.DiffSum2nd3rd,
DiffSum = trinucDiff.DiffSum
});
}
}
else if (item.SeqType == "CCDS" || item.SeqType == "CJCDS")
{
GBSequenceComp gbSequenceComp = new GBSequenceComp(subSeq);
subSeq = gbSequenceComp.SeqComp();
Console.Write($"{subSeq.Substring(0, 10)}...{subSeq.Substring(subSeq.Length - 10)}\t");
TrinucCalc trinucCalc = new TrinucCalc();
TrinucDiff trinucDiff = trinucCalc.Calculation(item.SeqType, subSeq);
Console.WriteLine($"{ trinucDiff.DiffSum1st2nd.ToString("0.0000") }\t" +
$"{ trinucDiff.DiffSum1st3rd.ToString("0.0000") }\t" +
$"{ trinucDiff.DiffSum2nd3rd.ToString("0.0000") }\t" +
$"{ trinucDiff.DiffSum.ToString("0.0000") }");
triDiff.Add(new TrinucDiff {
SeqType = item.SeqType,
DiffSum1st2nd = trinucDiff.DiffSum1st2nd,
DiffSum1st3rd = trinucDiff.DiffSum1st3rd,
DiffSum2nd3rd = trinucDiff.DiffSum2nd3rd,
DiffSum = trinucDiff.DiffSum
});
}
}
else
{
Console.WriteLine("nnnnnnnnnn...nnnnnnnnnn\t0.0000");
}
}
Console.WriteLine();
string filePath = string.Empty;
StreamWriter FH;
filePath = "/home/alvydas/Oligonucleotides/TrinucCalc/cdsframe.dat";
FH = new StreamWriter(filePath);
foreach (var item in triDiff)
{
if (item.SeqType == "CDS" || item.SeqType == "CCDS")
{
Console.WriteLine($"{item.SeqType}\t{item.DiffSum1st2nd.ToString("0.0000")}\t{item.DiffSum1st3rd.ToString("0.0000")}\t{item.DiffSum2nd3rd.ToString("0.0000")}\t{item.DiffSum.ToString("0.0000")}");
FH.WriteLine($"{item.SeqType}\t{item.DiffSum1st2nd.ToString("0.0000")}\t{item.DiffSum1st3rd.ToString("0.0000")}\t{item.DiffSum2nd3rd.ToString("0.0000")}\t{item.DiffSum.ToString("0.0000")}");
}
}
FH.Close();
Console.WriteLine();
filePath = "/home/alvydas/Oligonucleotides/TrinucCalc/ncdsframe.dat";
FH = new StreamWriter(filePath);
foreach (var item in triDiff)
{
if (item.SeqType == "NCDS")
{
Console.WriteLine($"{item.SeqType}\t{item.DiffSum1st2nd.ToString("0.0000")}\t{item.DiffSum1st3rd.ToString("0.0000")}\t{item.DiffSum2nd3rd.ToString("0.0000")}\t{item.DiffSum.ToString("0.0000")}");
FH.WriteLine($"{item.SeqType}\t{item.DiffSum1st2nd.ToString("0.0000")}\t{item.DiffSum1st3rd.ToString("0.0000")}\t{item.DiffSum2nd3rd.ToString("0.0000")}\t{item.DiffSum.ToString("0.0000")}");
}
}
FH.Close();
string rpath = "/usr/bin/Rscript";
//string arg1 = "/home/alvydas/Oligonucleotides/TrinucCalc/R/cdsframe.dat";
//string arg2 = "/home/alvydas/Oligonucleotides/TrinucCalc/R/ncdsframe.dat";
string scriptpath = "/home/alvydas/Oligonucleotides/TrinucCalc/framefrq.R cdsframe.dat ncdsframe.dat";
string output = RScript.RunRScript(rpath, scriptpath);
}
public static void Main(string [] args)
{
//string fileName = "/home/alvydas/Oligonucleotides/Sequencies/Bacteria/Escherichia.coli.gb";
string fileName = "/home/alvydas/Oligonucleotides/Sequencies/Bacteria/Escherichia.coli.test.gb";
//string fileName = "/home/alvydas/Oligonucleotides/Sequencies/Bacteria/Escherichia.coli.short.gb";
//string fileName = "/home/alvydas/Oligonucleotides/Sequencies/Bacteria/Escherichia.coli.short1.gb";
//string fileName = "/home/alvydas/Oligonucleotides/Sequencies/Bacteria/Escherichia.coli.short2.gb";
//string fileName = "/home/alvydas/Oligonucleotides/Sequencies/Bacteria/Escherichia.coli.shuffle.test.gb";
Console.WriteLine($"Calc seq cds trinucs for { Path.GetFileName (fileName) }");
GBSequence gbSequence = new GBSequence(fileName);
var gbSeq = gbSequence.GbSeq();
GBFeatures gbFeatures = new GBFeatures(fileName);
var gbCds = gbFeatures.Cds();
GBSeparation gBSeparation = new GBSeparation(gbSeq, gbCds);
gBSeparation.GBSeq();
TrinucDiff trinucDiff = new TrinucDiff();
Console.WriteLine($"Seq length {gbSeq.Seq.Length}\n{ gbSeq.Seq.Substring (0, 30)}....{ gbSeq.Seq.Substring (gbSeq.Seq.Length - 30)}\n");
int randFragLength = 1;
foreach (var item in gbCds)
{
if (item.CdsEnd - item.CdsStart >= 60)
{
TrinucCalc trinucCalc = new TrinucCalc();
double triDiff = trinucCalc.Calculation("CDS", item.CdsSeq);
Console.Write($"CDS\t{item.CdsStart}\t{item.CdsEnd}\t{item.CdsSeq.Substring (0, 9)}...{item.CdsSeq.Substring (item.CdsSeq.Length - 9)}\t");
Console.Write($"{trinucDiff.DiffSum1st2nd.ToString("0.0000")}\t{trinucDiff.DiffSum1st3rd.ToString ("0.0000")}\t{trinucDiff.DiffSum2nd3rd.ToString ("0.0000")}\t");
Console.WriteLine($"{triDiff.ToString("0.0000")}");
for (int i = 0; i < 10; i++)
{
RandomSeq randomSeq = new RandomSeq(item.CdsSeq);
string randSeq = randomSeq.RandomSeqByFragment(randFragLength);
TrinucCalc randtrinucCalc = new TrinucCalc();
triDiff = randtrinucCalc.Calculation("RCDS", randSeq);
Console.Write($"RCDS\t{item.CompCdsStart}\t{item.CompCdsEnd}\t{randSeq.Substring (0, 9)}...{randSeq.Substring (randSeq.Length - 9)}\t");
Console.Write($"{trinucDiff.DiffSum1st2nd.ToString ("0.0000")}\t{trinucDiff.DiffSum1st3rd.ToString ("0.0000")}\t{trinucDiff.DiffSum2nd3rd.ToString ("0.0000")}\t");
Console.WriteLine($"{triDiff.ToString ("0.0000")}");
}
}
if (item.CompCdsEnd - item.CompCdsStart >= 60)
{
TrinucCalc trinucCalc = new TrinucCalc();
double triDiff = trinucCalc.Calculation("CCDS", item.CompCdsSeq);
Console.Write($"CCDS\t{item.CompCdsStart}\t{item.CompCdsEnd}\t{item.CompCdsSeq.Substring (0, 9)}...{item.CompCdsSeq.Substring (item.CompCdsSeq.Length - 9)}\t");
Console.Write($"{trinucDiff.DiffSum1st2nd.ToString ("0.0000")}\t{trinucDiff.DiffSum1st3rd.ToString ("0.0000")}\t{trinucDiff.DiffSum2nd3rd.ToString ("0.0000")}\t");
Console.WriteLine($"{triDiff.ToString ("0.0000")}");
for (int i = 0; i < 10; i++)
{
RandomSeq randomSeq = new RandomSeq(item.CompCdsSeq);
string randSeq = randomSeq.RandomSeqByFragment(randFragLength);
TrinucCalc randtrinucCalc = new TrinucCalc();
triDiff = randtrinucCalc.Calculation("RCCDS", randSeq);
Console.Write($"RCCDS\t{item.CdsStart}\t{item.CdsEnd}\t{randSeq.Substring (0, 9)}...{randSeq.Substring (randSeq.Length - 9)}\t");
Console.Write($"{trinucDiff.DiffSum1st2nd.ToString ("0.0000")}\t{trinucDiff.DiffSum1st3rd.ToString ("0.0000")}\t{trinucDiff.DiffSum2nd3rd.ToString ("0.0000")}\t");
Console.WriteLine($"{triDiff.ToString ("0.0000")}");
}
}
//PrintOut printOut = new PrintOut (gbSeq, gbCds, 30, trinucDiff);
//printOut.SeqPrint ();
//printOut.CdsPrint ();
//printOut.CompCdsPrint ();
}
}
public TrinucDiff Calculation(string seqType, string seq)
{
TrinucCalculation trinucCalculation = new TrinucCalculation();
TrinucDiff trinucDiff = new TrinucDiff();
int trinucCount = 0;
string seqCalc = "nnn" + seq + seq.Substring(0, 9);
string oligoFrag = string.Empty;
for (int i = 0; i < seqCalc.Length - 9; i++)
{
oligoFrag = seqCalc.Substring(i, 3);
if (Array.IndexOf(trinucArray.trinuc, oligoFrag) < 0)
{
trinucleotideMatch.Add(new TrinucleotideMatch {
TrinucPosition = i, TrinucMatch = oligoFrag
});
continue;
}
if (i % 3 == 0)
{
var index = Array.FindIndex(trinucCalculation.trinucleotides, row => row.Trinuc == oligoFrag);
trinucCalculation.trinucleotides [index].Trinuc1st += 1;
trinucCount += 1;
}
else if ((i - 1) % 3 == 0)
{
var index = Array.FindIndex(trinucCalculation.trinucleotides, row => row.Trinuc == oligoFrag);
trinucCalculation.trinucleotides [index].Trinuc2nd += 1;
}
else if ((i - 2) % 3 == 0)
{
var index = Array.FindIndex(trinucCalculation.trinucleotides, row => row.Trinuc == oligoFrag);
trinucCalculation.trinucleotides [index].Trinuc3rd += 1;
}
}
trinucDiff.SeqType = seqType;
foreach (var item in trinucCalculation.trinucleotides)
{
item.TrinucFrq1st = ( double )item.Trinuc1st / trinucCount;
item.TrinucFrq2nd = ( double )item.Trinuc2nd / trinucCount;
item.TrinucFrq3rd = ( double )item.Trinuc3rd / trinucCount;
item.TrinucFrqDiff1st2nd = Math.Abs(item.TrinucFrq1st - item.TrinucFrq2nd);
item.TrinucFrqDiff1st3rd = Math.Abs(item.TrinucFrq1st - item.TrinucFrq3rd);
item.TrinucFrqDiff2nd3rd = Math.Abs(item.TrinucFrq2nd - item.TrinucFrq3rd);
}
foreach (var item in trinucCalculation.trinucleotides)
{
trinucDiff.DiffSum1st2nd += item.TrinucFrqDiff1st2nd;
trinucDiff.DiffSum1st3rd += item.TrinucFrqDiff1st3rd;
trinucDiff.DiffSum2nd3rd += item.TrinucFrqDiff2nd3rd;
}
trinucDiff.DiffSum = trinucDiff.DiffSum1st2nd + trinucDiff.DiffSum1st3rd + trinucDiff.DiffSum2nd3rd;
return(trinucDiff);
}