private void MakeUniprotDictionary(uniprot db)
{
uniprotDB.Clear();
foreach (var item in db.entry)
{
foreach (var accession in item.accession)
{
uniprotDB.Add(accession, item);
}
}
}
/// <summary>
/// Setup all for data loaded from UniProt
/// </summary>
/// <param name="db"></param>
public void SetupFromUniProt(uniprot db)
{
Log("Building chain names from UniProt dataset");
var proteinDB = from entry in db.entry
let chains = entry.GetChains().ToArray()
from chain in chains
let chainIndex = Array.IndexOf(chains, chain)
select new ChainDescriptor
{
ChainId = string.Format("{0}-{1}", entry.accession[0], chainIndex),
Accession = entry.accession[0],
ChainIndex = chainIndex,
Description = chain.description,
LocationBegin = chain.GetFirstLocation(),
LocationEnd = chain.GetLastLocation(),
Length = chain.GetFeatureLength(),
Lineage = entry.organism.lineage.LastOrDefault()
};
ProteinDB = proteinDB.ToList();
}
private static void Sample(string cleavageSiteName, uniprot db, IEnumerable<Tuple<TagInfo, TagInfo>> cleavageSites, string windowString, Workspace workspace)
{
var array = windowString.Split('-');
Debug.Assert(array.Length == 2, "Window string dont have 2 numbers");
var windowLeft = int.Parse(array[0]);
var windowRight = int.Parse(array[1]);
var acc = workspace.CreateSamplingData(windowLeft + windowRight);
acc.Class = cleavageSiteName + "|" + windowString;
foreach (var cleavageSite in cleavageSites)
{
Debug.Assert(cleavageSite.Item1.Accession == cleavageSite.Item2.Accession);
var entry = db.entry.First(x => x.accession.Contains(cleavageSite.Item1.Accession));
var text = CleanText(entry.sequence.Value);
Console.Error.WriteLine("Sampling window {0}-{1} applied to {2}:{3}", windowLeft, windowRight, cleavageSite.Item1.Accession, cleavageSiteName);
SamplePositivesNegatives(acc, text, windowLeft, windowRight, cleavageSite);
}
var num = acc.Positives.Count + acc.Negatives.Count;
Console.Error.WriteLine("Collected {0} samples (p:{1} / n:{2})", num, acc.Positives.Count, acc.Negatives.Count);
}