/// <summary>
/// Access the web to resolve protein metadata not directly found in fasta file.
/// The fasta text importer will have left search hints in ProteinMetadata.
/// </summary>
/// <param name="progressMonitor"></param>
/// <param name="status"></param>
/// <param name="fastaImporter">object that accesses the web, or pretends to if in a test</param>
/// <param name="parseOnly">if true, attempt to parse protein metadata from descriptions but do not proceed to web access</param>
/// <param name="done">will return true if there is nothung more to look up</param>
/// <returns>true on success</returns>
public bool LookupProteinMetadata(IProgressMonitor progressMonitor, ref IProgressStatus status, WebEnabledFastaImporter fastaImporter, bool parseOnly, out bool done)
{
var unsearchedProteins = new List <ProteinSearchInfo>();
done = false;
// If we're here, it's because the background loader is done digesting and has moved on to protein metadata,
// or because the PeptideSettingsUI thread needs to have protein metadata resolved for uniqueness purposes before
// it can proceed. Either way, we should be working on a temp copy and be the only one needing write access, so get a lock now
using (ISession session = OpenWriteSession()) // We may update the protdb file with web search results
{
if (!UpdateProgressAndCheckForCancellation(progressMonitor, ref status, Resources.ProteomeDb_LookupProteinMetadata_looking_for_unresolved_protein_details, 0))
{
return(false);
}
// get a list of proteins with unresolved metadata websearches
var proteinNames = session.CreateCriteria(typeof(DbProteinName)).List <DbProteinName>().Where(x => x.WebSearchInfo.NeedsSearch()).ToList();
var proteinsToSearch =
proteinNames.Where(proteinName => (proteinName.GetProteinMetadata().GetPendingSearchTerm().Length > 0))
.ToList();
if (!UpdateProgressAndCheckForCancellation(progressMonitor, ref status, Resources.ProteomeDb_LookupProteinMetadata_looking_for_unresolved_protein_details, 0))
{
return(false);
}
// and a list of proteins which have never been considered for metadata search
var untaggedProteins = proteinNames.Where(proteinName => proteinName.WebSearchInfo.IsEmpty()).ToList();
foreach (var untaggedProtein in untaggedProteins)
{
untaggedProtein.SetWebSearchCompleted(); // by default take this out of consideration for next time
var metadata = untaggedProtein.GetProteinMetadata();
if (metadata.HasMissingMetadata())
{
var search = fastaImporter.ParseProteinMetaData(metadata);
if (search != null)
{
metadata = untaggedProtein.ChangeProteinMetadata(metadata.Merge(search)); // don't stomp name by accident
metadata = untaggedProtein.ChangeProteinMetadata(metadata.ChangeWebSearchInfo(search.WebSearchInfo));
}
}
if (metadata.NeedsSearch())
{
proteinsToSearch.Add(untaggedProtein); // add to the list of things to commit back to the db
}
}
// Get the lengths of the sequences without getting the sequences themselves, for best speed
var proteinIds = proteinsToSearch.Select(name => name.Protein.Id.Value).Distinct().ToArray();
var proteinLengths = new Dictionary <long, int>();
using (var cmd = session.Connection.CreateCommand())
{
string sql = @"SELECT Id, LENGTH(Sequence) AS SequenceLength FROM ProteomeDbProtein P";
if (proteinIds.Length < 1000)
{
sql += @" WHERE P.Id IN (" +
string.Join(@",", proteinIds) + @")";
}
cmd.CommandText = sql;
using (var reader = cmd.ExecuteReader())
{
while (reader.Read())
{
var id = reader.GetValue(0);
var len = reader.GetValue(1);
proteinLengths.Add(Convert.ToInt64(id), Convert.ToInt32(len));
if (proteinLengths.Count % 100 == 0) // Periodic cancellation check
{
if (!UpdateProgressAndCheckForCancellation(progressMonitor, ref status, Resources.ProteomeDb_LookupProteinMetadata_looking_for_unresolved_protein_details, 0))
{
return(false);
}
}
}
}
}
foreach (var p in proteinsToSearch)
{
int length;
proteinLengths.TryGetValue(p.Protein.Id.GetValueOrDefault(), out length);
unsearchedProteins.Add(new ProteinSearchInfo(p, length));
}
if (untaggedProteins.Any(untagged => !untagged.GetProteinMetadata().NeedsSearch())) // did any get set as unsearchable?
{
// Write back the ones that were formerly without search terms, but which now indicate no search is possible
using (var transaction = session.BeginTransaction())
{
foreach (var untagged in untaggedProteins.Where(untagged => !untagged.GetProteinMetadata().NeedsSearch()))
{
session.SaveOrUpdate(untagged); // update the metadata
}
transaction.Commit();
}
}
if (unsearchedProteins.Any() && !parseOnly)
{
int resultsCount = 0;
int unsearchedCount = unsearchedProteins.Count;
for (bool success = true; success;)
{
success = false; // Until we see at least one succeed this round
var results = new List <DbProteinName>();
if (progressMonitor.IsCanceled)
{
return(false);
}
// The "true" arg means "do just one batch then return"
foreach (var result in fastaImporter.DoWebserviceLookup(unsearchedProteins, progressMonitor, true))
{
if (result != null)
{
string message = string.Format(Resources.ProteomeDb_LookupProteinMetadata_Retrieving_details_for__0__proteins,
unsearchedProteins.Count);
// Make it clearer when web access is faked during testing
if (fastaImporter.IsAccessFaked)
{
message = @"FAKED: " + message;
}
if (!UpdateProgressAndCheckForCancellation(progressMonitor, ref status, message, 100 * resultsCount++ / unsearchedCount))
{
return(false);
}
success = true;
results.Add(result.ProteinDbInfo);
}
}
if (results.Any()) // save this batch
{
using (var transaction = session.BeginTransaction())
{
foreach (var result in results)
{
session.SaveOrUpdate(result);
}
transaction.Commit();
}
}
// Edit this list rather than rederive with database access
var hits = unsearchedProteins.Where(p => !p.GetProteinMetadata().NeedsSearch()).ToList();
foreach (var hit in hits)
{
unsearchedProteins.Remove(hit);
}
}
}
done = !unsearchedProteins.Any();
} // End writesession
return(true);
}
/// <summary>
/// Access the web to resolve protein metadata not directly found in fasta file.
/// The fasta text importer will have left search hints in ProteinMetadata.
/// </summary>
/// <param name="progressMonitor"></param>
/// <param name="fastaImporter">object that accesses the web, or pretends to if in a test</param>
/// <param name="polite">if true, don't try to resolve everything in one go, assume we can come back later</param>
/// <returns>true on success</returns>
public bool LookupProteinMetadata(ProgressMonitor progressMonitor, WebEnabledFastaImporter fastaImporter, bool polite = false)
{
var unsearchedProteins = new List <ProteinSearchInfo>();
List <DbProteinName> untaggedProteins;
using (ISession session = OpenSession())
{
if (!progressMonitor.Invoke(Resources.ProteomeDb_LookupProteinMetadata_looking_for_unresolved_protein_details, 0))
{
return(false);
}
// get a list of proteins with unresolved metadata websearches
var proteinNames = session.CreateCriteria(typeof(DbProteinName)).List <DbProteinName>();
var proteinsToSearch =
proteinNames.Where(proteinName => (proteinName.GetProteinMetadata().GetPendingSearchTerm().Length > 0))
.ToList();
// and a list of proteins which have never been considered for metadata search
untaggedProteins =
proteinNames.Where(proteinName => proteinName.WebSearchInfo.IsEmpty()).ToList();
foreach (var untaggedProtein in untaggedProteins)
{
untaggedProtein.SetWebSearchCompleted(); // by default take this out of consideration for next time
var metadata = untaggedProtein.GetProteinMetadata();
if (metadata.HasMissingMetadata())
{
var search = fastaImporter.ParseProteinMetaData(metadata);
if (search != null)
{
metadata = untaggedProtein.ChangeProteinMetadata(metadata.Merge(search)); // don't stomp name by accident
metadata = untaggedProtein.ChangeProteinMetadata(metadata.ChangeWebSearchInfo(search.WebSearchInfo));
}
}
if (metadata.NeedsSearch())
{
proteinsToSearch.Add(untaggedProtein); // add to the list of things to commit back to the db
}
}
// Get the lengths of the sequences without getting the sequences themselves, for best speed
var proteinIds = proteinsToSearch.Select(name => name.Protein.Id.Value).Distinct().ToArray();
var proteinLengths = new Dictionary <long, int>();
using (var cmd = session.Connection.CreateCommand())
{
string sql = "SELECT Id, LENGTH(Sequence) AS SequenceLength FROM ProteomeDbProtein P"; // Not L10N
if (proteinIds.Length < 1000)
{
sql += " WHERE P.Id IN (" + // Not L10N
string.Join(",", proteinIds) + ")"; // Not L10N
}
cmd.CommandText = sql;
using (var reader = cmd.ExecuteReader())
{
while (reader.Read())
{
var id = reader.GetValue(0);
var len = reader.GetValue(1);
proteinLengths.Add(Convert.ToInt64(id), Convert.ToInt32(len));
}
}
}
foreach (var p in proteinsToSearch)
{
int length;
proteinLengths.TryGetValue(p.Protein.Id.GetValueOrDefault(), out length);
unsearchedProteins.Add(new ProteinSearchInfo(p, length));
}
}
if (untaggedProteins.Any(untagged => !untagged.GetProteinMetadata().NeedsSearch())) // did any get set as unsearchable?
{
// Write back the ones that were formerly without search terms, but which now indicate no search is possible
using (ISession session = OpenWriteSession())
{
using (var transaction = session.BeginTransaction())
{
foreach (var untagged in untaggedProteins.Where(untagged => !untagged.GetProteinMetadata().NeedsSearch()))
{
session.SaveOrUpdate(untagged); // update the metadata
}
transaction.Commit();
}
}
}
if (unsearchedProteins.Any())
{
int resultsCount = 0;
int unsearchedCount = unsearchedProteins.Count;
for (bool success = true; success;)
{
success = false; // Until we see at least one succeed this round
var results = new List <DbProteinName>();
// The "true" arg means "do just one batch then return"
foreach (var result in fastaImporter.DoWebserviceLookup(unsearchedProteins, null, true))
{
if (result != null)
{
if (
!progressMonitor.Invoke(
string.Format(
Resources.ProteomeDb_LookupProteinMetadata_Retrieving_details_for__0__proteins,
unsearchedProteins.Count), 100 * resultsCount++ / unsearchedCount))
{
return(false);
}
success = true;
results.Add(result.ProteinDbInfo);
}
}
if (results.Any()) // save this batch
{
using (var session = OpenWriteSession())
{
using (var transaction = session.BeginTransaction())
{
foreach (var result in results)
{
session.SaveOrUpdate(result);
}
transaction.Commit();
session.Close();
}
}
}
// Edit this list rather than rederive with database access
var hits = unsearchedProteins.Where(p => !p.GetProteinMetadata().NeedsSearch()).ToList();
foreach (var hit in hits)
{
unsearchedProteins.Remove(hit);
}
}
}
return(true);
}
