public UnparsedFile ReadFile(string [] lines)
{
var unparsedFile = new UnparsedFile();
foreach (string line in lines)
{
if (line.StartsWith("S"))
{
unparsedFile.CommentLineFromFile = line;
}
else if (line.StartsWith("f"))
{
unparsedFile.FamilyLinesFromFile.Add(line);
}
else if (line.StartsWith("a"))
{
unparsedFile.AccessionLinesFromFile.Add(line);
}
else if (line.StartsWith(">") || line.StartsWith("A") || line.StartsWith("T") || line.StartsWith("C") || line.StartsWith("G"))
{
unparsedFile.FastaLinesFromFile.Add(line);
}
}
return(unparsedFile);
}
public ParsedFile ParseFile(string fileName, UnparsedFile unparsedFile)
{
var partiallyParsedFileC = ParseCommentLine(fileName, unparsedFile);
var partiallyParsedFileCF = ParseFamilySection(partiallyParsedFileC, unparsedFile);
var partiallyParsedFileCFA = ParseAccessionSection(partiallyParsedFileCF, unparsedFile);
var parsedFile = ParseFastaSection(partiallyParsedFileCFA, unparsedFile);
var finishedFile = CreateDbEntry(parsedFile);
return(finishedFile);
}
public ParsedFile ParseAccessionSection(ParsedFile partiallyParsedFileCF, UnparsedFile unparsedFile)
{
int i = 1;
string[] temp;
foreach (string line in unparsedFile.AccessionLinesFromFile)
{
temp = line.Split(new char[] { ';' });
var accessionSection = ParseAccessionSectionLine(temp, i, unparsedFile.Sra);
partiallyParsedFileCF.AccessionSections.Add(accessionSection);
i++;
}
var partiallyParsedFileCFA = partiallyParsedFileCF;
return(partiallyParsedFileCFA);
}
public ParsedFile ParseFamilySection(ParsedFile partiallyParsedFileC, UnparsedFile unparsedFile)
{
int i = 1;
string[] temp;
foreach (string line in unparsedFile.FamilyLinesFromFile)
{
temp = line.Split(new char[] { ';' });
var familySection = ParseFamilySectionLine(temp, i, unparsedFile.Sra);
partiallyParsedFileC.FamilySections.Add(familySection);
i++;
}
var partiallyParsedFileCF = partiallyParsedFileC;
return(partiallyParsedFileCF);
}
public ParsedFile ParseCommentLine(string fileName, UnparsedFile unparsedFile)
{
string[] split = unparsedFile.CommentLineFromFile.Split(new char[] { ',' });
string[] sra = split[0].Split(new char[] { '=' });
string[] gen = split[1].Split(new char[] { '=' });
string[] date = split[2].Split(new char[] { '=' });
unparsedFile.Sra = sra[2];
var run = new Run();
run.FileName = fileName;
run.Sra = sra[2];
run.Genome = gen[1];
run.Date = date[1];
var partiallyParsedFile = new ParsedFile();
partiallyParsedFile.Run = run;
return(partiallyParsedFile);
}
public ParsedFile ParseFastaSection(ParsedFile partiallyParsedFileCFA, UnparsedFile unparsedFile)
{
bool first = true;
bool second = false;
int i = 1;
int j = 0;
List <string> firstLine = new List <string>();
List <string> secondLine = new List <string>();
List <int> lineNumber = new List <int>();
foreach (string line in unparsedFile.FastaLinesFromFile)
{
if (first)
{
firstLine.Add(line);
first = !first;
}
if (second)
{
secondLine.Add(line);
lineNumber.Add(i);
i++;
first = !first;
partiallyParsedFileCFA.FastaSections.Add(new FastaSection
{
FastaSectionLineId = lineNumber[j],
Sra = unparsedFile.Sra,
SequenceId = firstLine[j],
Sequence = secondLine[j]
});
j++;
}
second = !second;
}
var parsedFile = partiallyParsedFileCFA;
return(parsedFile);
}
