public void AddFastaFile(StreamReader reader, ProgressMonitor progressMonitor)
{
Dictionary<string, ProtIdNames> proteinIds = new Dictionary<string, ProtIdNames>();
using (ISession session = OpenWriteSession())
{
foreach (DbProtein protein in session.CreateCriteria(typeof(DbProtein)).List())
{
if (protein.Id.HasValue)
proteinIds.Add(protein.Sequence, new ProtIdNames(protein.Id.Value, protein.Names));
}
int proteinCount = 0;
using (var transaction = session.BeginTransaction())
using (IDbCommand insertProtein = session.Connection.CreateCommand())
using (IDbCommand insertName = session.Connection.CreateCommand())
{
WebEnabledFastaImporter fastaImporter = new WebEnabledFastaImporter(new WebEnabledFastaImporter.DelayedWebSearchProvider()); // just parse, no search for now
insertProtein.CommandText =
"INSERT INTO ProteomeDbProtein (Version, Sequence) Values (1,?);select last_insert_rowid();"; // Not L10N
insertProtein.Parameters.Add(new SQLiteParameter());
insertName.CommandText =
"INSERT INTO ProteomeDbProteinName (Version, Protein, IsPrimary, Name, Description, PreferredName, Accession, Gene, Species, WebSearchStatus) Values(1,?,?,?,?,?,?,?,?,?)"; // Not L10N
insertName.Parameters.Add(new SQLiteParameter()); // Id
insertName.Parameters.Add(new SQLiteParameter()); // IsPrimary
insertName.Parameters.Add(new SQLiteParameter()); // Name
insertName.Parameters.Add(new SQLiteParameter()); // Description
insertName.Parameters.Add(new SQLiteParameter()); // PreferredName
insertName.Parameters.Add(new SQLiteParameter()); // Accession
insertName.Parameters.Add(new SQLiteParameter()); // Gene
insertName.Parameters.Add(new SQLiteParameter()); // Species
insertName.Parameters.Add(new SQLiteParameter()); // WebSearchInfo
foreach (DbProtein protein in fastaImporter.Import(reader))
{
int iProgress = (int)(reader.BaseStream.Position * 100 / (reader.BaseStream.Length + 1));
if (!progressMonitor.Invoke(string.Format(Resources.ProteomeDb_AddFastaFile_Added__0__proteins,proteinCount), iProgress))
{
return;
}
bool existingProtein = false;
ProtIdNames proteinIdNames;
if (proteinIds.TryGetValue(protein.Sequence, out proteinIdNames))
{
existingProtein = true;
}
else
{
((SQLiteParameter)insertProtein.Parameters[0]).Value = protein.Sequence;
proteinIdNames = new ProtIdNames(Convert.ToInt64(insertProtein.ExecuteScalar()), new DbProteinName[0]);
proteinIds.Add(protein.Sequence, proteinIdNames);
proteinCount++;
}
foreach (var proteinName in protein.Names)
{
// Skip any names that already exist
if (proteinIdNames.Names.Any(dbProteinName => Equals(dbProteinName.Name, proteinName.Name)))
continue;
try
{
((SQLiteParameter)insertName.Parameters[0]).Value = proteinIdNames.Id;
((SQLiteParameter)insertName.Parameters[1]).Value = proteinName.IsPrimary && !existingProtein;
((SQLiteParameter)insertName.Parameters[2]).Value = proteinName.Name;
((SQLiteParameter)insertName.Parameters[3]).Value = proteinName.Description;
((SQLiteParameter)insertName.Parameters[4]).Value = proteinName.PreferredName;
((SQLiteParameter)insertName.Parameters[5]).Value = proteinName.Accession;
((SQLiteParameter)insertName.Parameters[6]).Value = proteinName.Gene;
((SQLiteParameter)insertName.Parameters[7]).Value = proteinName.Species;
((SQLiteParameter)insertName.Parameters[8]).Value = proteinName.WebSearchStatus; // represent as a string for ease of serialization
insertName.ExecuteNonQuery();
}
catch (Exception exception)
{
Console.Out.WriteLine(exception);
}
}
}
if (!progressMonitor.Invoke(Resources.ProteomeDb_AddFastaFile_Saving_changes, 99))
{
return;
}
transaction.Commit();
}
AnalyzeDb(session);
progressMonitor.Invoke(
string.Format(Resources.ProteomeDb_AddFastaFile_Finished_importing__0__proteins, proteinCount), 100);
}
}
private void DoWork()
{
try
{
// just do the basic name+description parsing, no regex or web access - we don't use extended metadata here
var fastaImporter = new WebEnabledFastaImporter(new WebEnabledFastaImporter.FakeWebSearchProvider());
var proteins = new Dictionary<string, ProteinData>();
foreach (var path in FastaFilePaths)
{
var statusText = "Reading FASTA file " + Path.GetFileName(path);
var fileInfo = new FileInfo(path);
var reader = File.OpenText(path);
foreach (var protein in fastaImporter.Import(File.OpenText(path)))
{
if (!UpdateProgress(statusText, (int) (reader.BaseStream.Position * 100 / fileInfo.Length)))
{
return;
}
ProteinData proteinData;
if (!proteins.TryGetValue(protein.Sequence, out proteinData))
{
proteinData = new ProteinData(protein.Sequence);
proteins.Add(protein.Sequence, proteinData);
}
foreach (var name in protein.Names)
{
proteinData.AddName(name.Name, name.Description);
}
}
}
UpdatePeptides(proteins);
if (!IsDisposed)
{
BeginInvoke(new Action(Close));
}
}
finally
{
lock(this)
{
_running = false;
Monitor.PulseAll(this);
}
}
}