public List <SearchSiteCollection> FetchSearchSiteData(GeneCollection geneCollection, string originSequence, string[] siteArray)
{
Dictionary <string, string> iupacTable = CreateIupacTable();
List <SearchSiteCollection> finalSiteList = new List <SearchSiteCollection>();
List <string> nonMatchedSites = new List <string>();
string upperOriginSequence = originSequence.ToUpper();
foreach (string site in siteArray)
{
SearchSiteCollection searchSiteCollection = new SearchSiteCollection();
string[] nucleotideSites = site.Split();
List <string> targetSequences = new List <string>();
bool isMatch = false;
foreach (char nucleotideSite in site)
{
string nucSite = nucleotideSite.ToString();
if (iupacTable.ContainsKey(nucSite))
{
String newNucleotides = iupacTable[nucSite];
String regex = "[" + newNucleotides + "]";
foreach (char newNuc in newNucleotides)
{
string nuc = newNuc.ToString();
string targetSequence = site.Replace(nucSite, nuc);
targetSequences.Add(targetSequence);
}
searchSiteCollection.searchSiteRegexMessage = "site search: " + nucSite + " (regex: " + nucSite.Replace(nucSite, regex) + ")";
}
}
foreach (String targetSequence in targetSequences)
{
int index = 0;
while (index >= 0 && index != upperOriginSequence.Length)
{
index++;
index = upperOriginSequence.IndexOf(targetSequence, index);
String strIndex = "" + (index + 1);
if (index != -1)
{
String geneName = "INTERGENIC";
foreach (Gene gene in geneCollection.collection)
{
if (index >= gene.StartCoordinate && index <= gene.EndCoordinate)
{
//Index + 1 because coding index starts at zero and sequence index starts at 1.
if (gene.ID != "")
{
geneName = gene.ID;
}
else
{
geneName = gene.LocusTag;
}
isMatch = true;
}
}
SearchSite searchSite = new SearchSite(strIndex, targetSequence, geneName);
if (searchSiteCollection.collection.Any())
{
bool newEntry = true;
foreach (SearchSite siteEntry in searchSiteCollection.collection)
{
if (siteEntry.StartPosition.Equals(strIndex))
{
if (siteEntry.Site.Equals("INTERGENIC") && !geneName.Equals("INTERGENIC"))
{
siteEntry.GeneName = geneName;
newEntry = false;
break;
}
}
}
if (newEntry)
{
searchSiteCollection.AddSearchSite(searchSite);
}
}
else
{
searchSiteCollection.AddSearchSite(searchSite);
}
}
}
}
finalSiteList.Add(searchSiteCollection);
if (!isMatch)
{
nonMatchedSites.Add(site);
}
}
//If all site entries did not match:
if (nonMatchedSites.Any())
{
//Display list of nucleotide site entries that did not match.
String message = "";
foreach (String site in nonMatchedSites)
{
if (message == "")
{
message += "Some of the provided entries could not be found. Here is the list of nucleotide sites; ";
message += site;
}
else
{
message += ", " + site;
}
}
Console.WriteLine("\n" + message);
}
return(finalSiteList);
}
public void AddSearchSite(SearchSite site)
{
collection.Add(site);
}