public Digestion GetDigestion(String name)
{
DbDigestion digestion = GetDbDigestion(name);
if (digestion == null)
{
return(null);
}
return(new Digestion(this, digestion));
}
public Digestion Digest(IProtease protease, String name, String description, ProgressMonitor progressMonitor)
{
DbOrganism organism;
DbDigestion digestion;
List <DbProtein> proteins;
using (ISession session = ProteomeDb.OpenWriteSession())
{
organism = GetEntity(session);
session.BeginTransaction();
digestion = new DbDigestion
{
Name = name,
Description = description,
Organism = organism,
MaxMissedCleavages = protease.MaxMissedCleavages
};
session.Save(digestion);
if (!progressMonitor.Invoke("Listing proteins", 0))
{
return(null);
}
proteins = new List <DbProtein>(organism.Proteins);
Dictionary <String, long> digestedPeptideIds
= new Dictionary <string, long>();
const String sqlPeptide =
"INSERT INTO ProteomeDbDigestedPeptide (Digestion, MissedCleavages, Sequence, Version) VALUES(@Digestion,@MissedCleavages,@Sequence,1);select last_insert_rowid();";
var commandPeptide = session.Connection.CreateCommand();
commandPeptide.CommandText = sqlPeptide;
commandPeptide.Parameters.Add(new SQLiteParameter("@Digestion"));
commandPeptide.Parameters.Add(new SQLiteParameter("@MissedCleavages"));
commandPeptide.Parameters.Add(new SQLiteParameter("@Sequence"));
const String sqlPeptideProtein =
"INSERT INTO ProteomeDbDigestedPeptideProtein (StartIndex, Peptide, Protein, Version) VALUES(?,?,?,1);";
var commandProtein = session.Connection.CreateCommand();
commandProtein.CommandText = sqlPeptideProtein;
commandProtein.Parameters.Add(new SQLiteParameter("@StartIndex"));
commandProtein.Parameters.Add(new SQLiteParameter("@Peptide"));
commandProtein.Parameters.Add(new SQLiteParameter("@Protein"));
for (int i = 0; i < proteins.Count; i++)
{
if (!progressMonitor.Invoke("Digesting " + proteins.Count
+ " proteins", 100 * i / proteins.Count))
{
return(null);
}
Protein protein = new Protein(this, proteins[i]);
foreach (DigestedPeptide digestedPeptide in protease.Digest(protein))
{
if (digestedPeptide.Sequence.Length > MAX_PEPTIDE_LENGTH)
{
continue;
}
long digestedPeptideId;
if (!digestedPeptideIds.TryGetValue(digestedPeptide.Sequence, out digestedPeptideId))
{
((SQLiteParameter)commandPeptide.Parameters[0]).Value = digestion.Id;
((SQLiteParameter)commandPeptide.Parameters[1]).Value = digestedPeptide.MissedCleavages;
((SQLiteParameter)commandPeptide.Parameters[2]).Value = digestedPeptide.Sequence;
digestedPeptideId = Convert.ToInt64(commandPeptide.ExecuteScalar());
digestedPeptideIds.Add(digestedPeptide.Sequence, digestedPeptideId);
}
((SQLiteParameter)commandProtein.Parameters[0]).Value = digestedPeptide.Index;
((SQLiteParameter)commandProtein.Parameters[1]).Value = digestedPeptideId;
((SQLiteParameter)commandProtein.Parameters[2]).Value = proteins[i].Id;
commandProtein.ExecuteNonQuery();
}
}
if (!progressMonitor.Invoke("Committing transaction", 99))
{
return(null);
}
session.Transaction.Commit();
progressMonitor.Invoke(
"Digested " + proteins.Count + " proteins into " + digestedPeptideIds.Count + " unique peptides",
100);
return(new Digestion(this, digestion));
}
}
public Digestion Digest(IProtease protease, ProgressMonitor progressMonitor)
{
using (ISession session = OpenWriteSession())
{
DbDigestion dbDigestion = GetDbDigestion(protease.Name);
HashSet <string> existingSequences = new HashSet <string>();
using (var transaction = session.BeginTransaction())
{
if (dbDigestion != null)
{
if (dbDigestion.MaxSequenceLength >= MAX_SEQUENCE_LENGTH)
{
return(new Digestion(this, dbDigestion));
}
if (!progressMonitor.Invoke(Resources.ProteomeDb_Digest_Listing_existing_peptides, 0))
{
return(null);
}
IQuery query = session.CreateQuery("SELECT P.Sequence FROM " // Not L10N
+ typeof(DbDigestedPeptide) + " P WHERE P.Digestion = :Digestion") // Not L10N
.SetParameter("Digestion", dbDigestion); // Not L10N
List <String> listSequences = new List <string>();
query.List(listSequences);
existingSequences.UnionWith(listSequences);
dbDigestion.MaxSequenceLength = MAX_SEQUENCE_LENGTH;
session.Update(dbDigestion);
}
else
{
dbDigestion = new DbDigestion
{
Name = protease.Name,
MinSequenceLength = MIN_SEQUENCE_LENGTH,
MaxSequenceLength = MAX_SEQUENCE_LENGTH,
};
session.Save(dbDigestion);
}
if (!progressMonitor.Invoke(Resources.ProteomeDb_Digest_Listing_proteins, 0))
{
return(null);
}
List <DbProtein> proteins = new List <DbProtein>();
session.CreateCriteria(typeof(DbProtein)).List(proteins);
Dictionary <String, long> digestedPeptideIds
= new Dictionary <string, long>();
const String sqlPeptide =
"INSERT INTO ProteomeDbDigestedPeptide (Digestion, Sequence) VALUES(?,?);select last_insert_rowid();"; // Not L10N
using (var commandPeptide = session.Connection.CreateCommand())
using (var commandProtein = session.Connection.CreateCommand())
{
commandPeptide.CommandText = sqlPeptide;
commandPeptide.Parameters.Add(new SQLiteParameter());
commandPeptide.Parameters.Add(new SQLiteParameter());
const String sqlPeptideProtein =
"INSERT INTO ProteomeDbDigestedPeptideProtein (Peptide, Protein) VALUES(?,?);"; // Not L10N
commandProtein.CommandText = sqlPeptideProtein;
commandProtein.Parameters.Add(new SQLiteParameter());
commandProtein.Parameters.Add(new SQLiteParameter());
commandProtein.Parameters.Add(new SQLiteParameter());
for (int i = 0; i < proteins.Count; i++)
{
var proteinSequences = new HashSet <string>();
if (!progressMonitor.Invoke(string.Format(Resources.ProteomeDb_Digest_Digesting__0__proteins, proteins.Count), 100 * i / proteins.Count))
{
return(null);
}
Protein protein = new Protein(ProteomeDbPath, proteins[i]);
foreach (DigestedPeptide digestedPeptide in protease.Digest(protein))
{
if (digestedPeptide.Sequence.Length < dbDigestion.MinSequenceLength)
{
continue;
}
String truncatedSequence = digestedPeptide.Sequence.Substring(
0, Math.Min(digestedPeptide.Sequence.Length, dbDigestion.MaxSequenceLength));
if (existingSequences.Contains(truncatedSequence))
{
continue;
}
if (proteinSequences.Contains(truncatedSequence))
{
continue;
}
proteinSequences.Add(truncatedSequence);
long digestedPeptideId;
if (!digestedPeptideIds.TryGetValue(truncatedSequence, out digestedPeptideId))
{
((SQLiteParameter)commandPeptide.Parameters[0]).Value = dbDigestion.Id;
((SQLiteParameter)commandPeptide.Parameters[1]).Value = truncatedSequence;
digestedPeptideId = Convert.ToInt64(commandPeptide.ExecuteScalar());
digestedPeptideIds.Add(truncatedSequence, digestedPeptideId);
}
((SQLiteParameter)commandProtein.Parameters[0]).Value = digestedPeptideId;
((SQLiteParameter)commandProtein.Parameters[1]).Value = protein.Id;
commandProtein.ExecuteNonQuery();
}
}
}
if (!progressMonitor.Invoke(Resources.ProteomeDb_AddFastaFile_Saving_changes, 99))
{
return(null);
}
transaction.Commit();
AnalyzeDb(session);
progressMonitor.Invoke(
string.Format(Resources.ProteomeDb_Digest_Digested__0__proteins_into__1__unique_peptides,
proteins.Count, digestedPeptideIds.Count),
100);
}
return(new Digestion(this, dbDigestion));
}
}
public Digestion Digest(IProtease protease, ProgressMonitor progressMonitor)
{
using (ISession session = OpenWriteSession())
{
DbDigestion dbDigestion = GetDbDigestion(protease.Name);
HashSet<string> existingSequences = new HashSet<string>();
using (var transaction = session.BeginTransaction())
{
if (dbDigestion != null)
{
if (dbDigestion.MaxSequenceLength >= MAX_SEQUENCE_LENGTH)
{
return new Digestion(this, dbDigestion);
}
if (!progressMonitor.Invoke(Resources.ProteomeDb_Digest_Listing_existing_peptides, 0))
{
return null;
}
IQuery query = session.CreateQuery("SELECT P.Sequence FROM " // Not L10N
+ typeof(DbDigestedPeptide) + " P WHERE P.Digestion = :Digestion") // Not L10N
.SetParameter("Digestion", dbDigestion); // Not L10N
List<String> listSequences = new List<string>();
query.List(listSequences);
existingSequences.UnionWith(listSequences);
dbDigestion.MaxSequenceLength = MAX_SEQUENCE_LENGTH;
session.Update(dbDigestion);
}
else
{
dbDigestion = new DbDigestion
{
Name = protease.Name,
MinSequenceLength = MIN_SEQUENCE_LENGTH,
MaxSequenceLength = MAX_SEQUENCE_LENGTH,
};
session.Save(dbDigestion);
}
if (!progressMonitor.Invoke(Resources.ProteomeDb_Digest_Listing_proteins, 0))
{
return null;
}
List<DbProtein> proteins = new List<DbProtein>();
session.CreateCriteria(typeof(DbProtein)).List(proteins);
Dictionary<String, long> digestedPeptideIds
= new Dictionary<string, long>();
const String sqlPeptide =
"INSERT INTO ProteomeDbDigestedPeptide (Digestion, Sequence) VALUES(?,?);select last_insert_rowid();"; // Not L10N
using (var commandPeptide = session.Connection.CreateCommand())
using (var commandProtein = session.Connection.CreateCommand())
{
commandPeptide.CommandText = sqlPeptide;
commandPeptide.Parameters.Add(new SQLiteParameter());
commandPeptide.Parameters.Add(new SQLiteParameter());
const String sqlPeptideProtein =
"INSERT INTO ProteomeDbDigestedPeptideProtein (Peptide, Protein) VALUES(?,?);"; // Not L10N
commandProtein.CommandText = sqlPeptideProtein;
commandProtein.Parameters.Add(new SQLiteParameter());
commandProtein.Parameters.Add(new SQLiteParameter());
commandProtein.Parameters.Add(new SQLiteParameter());
for (int i = 0; i < proteins.Count; i++)
{
var proteinSequences = new HashSet<string>();
if (!progressMonitor.Invoke(string.Format(Resources.ProteomeDb_Digest_Digesting__0__proteins,proteins.Count), 100 * i / proteins.Count))
{
return null;
}
Protein protein = new Protein(ProteomeDbPath, proteins[i]);
foreach (DigestedPeptide digestedPeptide in protease.Digest(protein))
{
if (digestedPeptide.Sequence.Length < dbDigestion.MinSequenceLength)
{
continue;
}
String truncatedSequence = digestedPeptide.Sequence.Substring(
0, Math.Min(digestedPeptide.Sequence.Length, dbDigestion.MaxSequenceLength));
if (existingSequences.Contains(truncatedSequence))
{
continue;
}
if (proteinSequences.Contains(truncatedSequence))
{
continue;
}
proteinSequences.Add(truncatedSequence);
long digestedPeptideId;
if (!digestedPeptideIds.TryGetValue(truncatedSequence, out digestedPeptideId))
{
((SQLiteParameter)commandPeptide.Parameters[0]).Value = dbDigestion.Id;
((SQLiteParameter)commandPeptide.Parameters[1]).Value = truncatedSequence;
digestedPeptideId = Convert.ToInt64(commandPeptide.ExecuteScalar());
digestedPeptideIds.Add(truncatedSequence, digestedPeptideId);
}
((SQLiteParameter)commandProtein.Parameters[0]).Value = digestedPeptideId;
((SQLiteParameter)commandProtein.Parameters[1]).Value = protein.Id;
commandProtein.ExecuteNonQuery();
}
}
}
if (!progressMonitor.Invoke(Resources.ProteomeDb_AddFastaFile_Saving_changes, 99))
{
return null;
}
transaction.Commit();
AnalyzeDb(session);
progressMonitor.Invoke(
string.Format(Resources.ProteomeDb_Digest_Digested__0__proteins_into__1__unique_peptides,
proteins.Count, digestedPeptideIds.Count),
100);
}
return new Digestion(this, dbDigestion);
}
}
public Digestion Digest(IProtease protease, int maxMissedCleavages, IProgressMonitor progressMonitor, ref IProgressStatus status, bool delayDbIndexing = false)
{
using (ISession session = OpenWriteSession())
{
DbDigestion dbDigestion = GetDbDigestion(protease.Name, session);
HashSet <string> existingSequences; // TODO(bspratt) - the logic around this seems fishy, investigate. Probably never actually been used. Part of fix for issue #304, probably
if (dbDigestion != null)
{
if (dbDigestion.MaxSequenceLength >= MAX_SEQUENCE_LENGTH)
{
return(new Digestion(this, dbDigestion));
}
if (!UpdateProgressAndCheckForCancellation(progressMonitor, ref status, Resources.ProteomeDb_Digest_Listing_existing_peptides, 0))
{
return(null);
}
IQuery query = session.CreateQuery("SELECT P.Sequence FROM " // Not L10N
+ typeof(DbDigestedPeptide) + " P WHERE P.Digestion = :Digestion") // Not L10N
.SetParameter("Digestion", dbDigestion); // Not L10N
List <String> listSequences = new List <string>();
query.List(listSequences);
existingSequences = new HashSet <string>(listSequences);
dbDigestion.MaxSequenceLength = MAX_SEQUENCE_LENGTH;
}
else
{
dbDigestion = new DbDigestion
{
Name = protease.Name,
MinSequenceLength = MIN_SEQUENCE_LENGTH,
MaxSequenceLength = MAX_SEQUENCE_LENGTH,
};
existingSequences = new HashSet <string>();
}
if (!UpdateProgressAndCheckForCancellation(progressMonitor, ref status, Resources.ProteomeDb_Digest_Listing_proteins, 0))
{
return(null);
}
var dbProteins = new List <DbProtein>();
session.CreateCriteria(typeof(DbProtein)).List(dbProteins);
// Digest the proteins
var proteinCount = dbProteins.Count;
if (proteinCount == 0)
{
return(null);
}
var proteinsList = new Protein[proteinCount];
var truncatedSequences = new HashSet <string> [proteinCount]; // One hashset of sequences for each protein of interest
const int N_DIGEST_THREADS = 16; // Arbitrary value - do a progress/canel check every nth protein
string message = string.Format(Resources.ProteomeDb_Digest_Digesting__0__proteins, proteinCount);
for (var i = 0; i < proteinCount; i += N_DIGEST_THREADS)
{
var endRange = Math.Min(proteinCount, i + N_DIGEST_THREADS);
if (!UpdateProgressAndCheckForCancellation(progressMonitor, ref status, message, 50 * endRange / proteinCount))
{
return(null);
}
for (int ii = i; ii < endRange; ii++)
{
var protein = new Protein(ProteomeDbPath, dbProteins[ii]);
proteinsList[ii] = protein;
}
Parallel.For(i, endRange, ii =>
{
var proteinSequences = new HashSet <string>(); // We only save the first dbDigestion.MaxSequenceLength characters of each peptide so collisions are likely
truncatedSequences[ii] = proteinSequences; // One hashset of sequences for each protein of interest
foreach (var digestedPeptide in protease.DigestSequence(proteinsList[ii].Sequence, maxMissedCleavages, null))
{
if (digestedPeptide.Sequence.Length < dbDigestion.MinSequenceLength)
{
continue;
}
var truncatedSequence = digestedPeptide.Sequence.Substring(
0, Math.Min(digestedPeptide.Sequence.Length, dbDigestion.MaxSequenceLength));
if (!existingSequences.Contains(truncatedSequence))
{
proteinSequences.Add(truncatedSequence);
}
}
});
}
// Now write to db
if (!UpdateProgressAndCheckForCancellation(progressMonitor, ref status, Resources.ProteomeDb_AddFastaFile_Saving_changes, 50))
{
return(null);
}
bool committed = true;
int digestedPeptideIdsCount;
try
{
using (var transaction = session.BeginTransaction())
{
session.SaveOrUpdate(dbDigestion);
Dictionary <String, long> digestedPeptideIds
= new Dictionary <string, long>();
const String sqlPeptide =
"INSERT INTO ProteomeDbDigestedPeptide (Digestion, Sequence) VALUES(?,?);select last_insert_rowid();"; // Not L10N
using (var commandPeptide = session.Connection.CreateCommand())
using (var commandProtein = session.Connection.CreateCommand())
{
commandPeptide.CommandText = sqlPeptide;
commandPeptide.Parameters.Add(new SQLiteParameter());
commandPeptide.Parameters.Add(new SQLiteParameter());
const String sqlPeptideProtein =
"INSERT INTO ProteomeDbDigestedPeptideProtein (Peptide, Protein) VALUES(?,?);"; // Not L10N
commandProtein.CommandText = sqlPeptideProtein;
commandProtein.Parameters.Add(new SQLiteParameter());
commandProtein.Parameters.Add(new SQLiteParameter());
commandProtein.Parameters.Add(new SQLiteParameter());
for (int i = 0; i < proteinCount; i++)
{
var protein = proteinsList[i];
if (!UpdateProgressAndCheckForCancellation(progressMonitor, ref status, message, 50 * (proteinCount + i) / proteinCount))
{
return(null);
}
foreach (var truncatedSequence in truncatedSequences[i])
{
long digestedPeptideId;
if (!digestedPeptideIds.TryGetValue(truncatedSequence, out digestedPeptideId))
{
((SQLiteParameter)commandPeptide.Parameters[0]).Value = dbDigestion.Id;
((SQLiteParameter)commandPeptide.Parameters[1]).Value = truncatedSequence;
digestedPeptideId = Convert.ToInt64(commandPeptide.ExecuteScalar());
digestedPeptideIds.Add(truncatedSequence, digestedPeptideId);
}
((SQLiteParameter)commandProtein.Parameters[0]).Value = digestedPeptideId;
((SQLiteParameter)commandProtein.Parameters[1]).Value = protein.Id;
commandProtein.ExecuteNonQuery();
}
}
}
try
{
transaction.Commit();
}
catch (Exception)
{
committed = false;
}
digestedPeptideIdsCount = digestedPeptideIds.Count;
}
}
catch (Exception)
{
if (!committed)
{
return(null); // Interrupted
}
else
{
throw;
}
}
if (committed && !delayDbIndexing)
{
AnalyzeDb(session); // This runs asynchronously, and interferes with writes
}
if (committed)
{
progressMonitor.UpdateProgress(new ProgressStatus(string.Format(Resources.ProteomeDb_Digest_Digested__0__proteins_into__1__unique_peptides, proteinCount, digestedPeptideIdsCount)).ChangePercentComplete(100));
}
return(committed ? new Digestion(this, dbDigestion) : null);
}
}
