public void AddFastaFile(string fastaFilePath)
{
String databasePath = textPath.Text;
Settings.Default.FastaDirectory = Path.GetDirectoryName(fastaFilePath);
int duplicateSequenceCount = 0;
using (var longWaitDlg = new LongWaitDlg {
ProgressValue = 0
})
{
try
{
longWaitDlg.PerformWork(this, 0, progressMonitor =>
{
ProteomeDb proteomeDb = File.Exists(databasePath)
? ProteomeDb.OpenProteomeDb(databasePath)
: ProteomeDb.CreateProteomeDb(databasePath);
using (proteomeDb)
{
using (var reader = File.OpenText(fastaFilePath))
{
IProgressStatus status = new ProgressStatus(longWaitDlg.Message);
proteomeDb.AddFastaFile(reader, progressMonitor, ref status, false, out duplicateSequenceCount);
}
}
});
}
catch (Exception x)
{
var message = TextUtil.LineSeparate(string.Format(Resources.BuildBackgroundProteomeDlg_AddFastaFile_An_error_occurred_attempting_to_add_the_FASTA_file__0__,
fastaFilePath), x.Message);
MessageDlg.ShowWithException(this, message, x);
return;
}
}
string path = Path.GetFileName(fastaFilePath);
if (path != null)
{
listboxFasta.Items.Add(path);
}
RefreshStatus();
if (duplicateSequenceCount > 0)
{
MessageDlg.Show(this, string.Format(Resources.BuildBackgroundProteomeDlg_AddFastaFile_The_added_file_included__0__repeated_protein_sequences__Their_names_were_added_as_aliases_to_ensure_the_protein_list_contains_only_one_copy_of_each_sequence_, duplicateSequenceCount));
}
}
public void AddFastaFile(string fastaFilePath)
{
String databasePath = textPath.Text;
Settings.Default.FastaDirectory = Path.GetDirectoryName(fastaFilePath);
using (var longWaitDlg = new LongWaitDlg {
ProgressValue = 0
})
{
var progressMonitor = new ProgressMonitor(longWaitDlg);
try
{
longWaitDlg.PerformWork(this, 0, () =>
{
ProteomeDb proteomeDb = File.Exists(databasePath)
? ProteomeDb.OpenProteomeDb(databasePath)
: ProteomeDb.CreateProteomeDb(databasePath);
using (proteomeDb)
{
using (var reader = File.OpenText(fastaFilePath))
{
proteomeDb.AddFastaFile(reader, progressMonitor.UpdateProgress);
}
}
});
}
catch (Exception x)
{
var message = TextUtil.LineSeparate(string.Format(Resources.BuildBackgroundProteomeDlg_AddFastaFile_An_error_occurred_attempting_to_add_the_FASTA_file__0__,
fastaFilePath), x.Message);
MessageDlg.ShowWithException(this, message, x);
return;
}
}
string path = Path.GetFileName(fastaFilePath);
if (path != null)
{
listboxFasta.Items.Add(path);
}
RefreshStatus();
}
public void TestProteomeDb()
{
using (var testFilesDir = new TestFilesDir(TestContext, ZIP_FILE))
{
string fastaPath = testFilesDir.GetTestPath("high_ipi.Human.20060111.fasta");
string protDbPath = testFilesDir.GetTestPath("test.protdb");
using (ProteomeDb proteomeDb = ProteomeDb.CreateProteomeDb(protDbPath))
{
IProgressStatus status = new ProgressStatus(string.Empty);
using (var reader = new StreamReader(fastaPath))
{
proteomeDb.AddFastaFile(reader, new SilentProgressMonitor(), ref status, true); // Delay indexing
}
// perform digestion
Digestion digestion = proteomeDb.GetDigestion();
var digestedProteins0 = digestion.GetProteinsWithSequence("EDGWVK");
Assert.IsTrue(digestedProteins0.Count >= 1);
}
}
}
public void TestProteomeDb()
{
using (var testFilesDir = new TestFilesDir(TestContext, ZIP_FILE))
{
string fastaPath = testFilesDir.GetTestPath("high_ipi.Human.20060111.fasta");
string protDbPath = testFilesDir.GetTestPath("test.protdb");
using (ProteomeDb proteomeDb = ProteomeDb.CreateProteomeDb(protDbPath))
{
Enzyme trypsin = EnzymeList.GetDefault();
using (var reader = new StreamReader(fastaPath))
{
proteomeDb.AddFastaFile(reader, (msg, progress) => true);
}
// perform digestion
proteomeDb.Digest(new ProteaseImpl(trypsin), (msg, progress) => true);
Digestion digestion = proteomeDb.GetDigestion(trypsin.Name);
var digestedProteins0 = digestion.GetProteinsWithSequencePrefix("EDGWVK", 100);
Assert.IsTrue(digestedProteins0.Count >= 1);
}
}
}
public void TestProteomeDb()
{
using (var testFilesDir = new TestFilesDir(TestContext, ZIP_FILE))
{
string fastaPath = testFilesDir.GetTestPath("high_ipi.Human.20060111.fasta");
string protDbPath = testFilesDir.GetTestPath("test.protdb");
using (ProteomeDb proteomeDb = ProteomeDb.CreateProteomeDb(protDbPath))
{
Enzyme trypsin = EnzymeList.GetDefault();
IProgressStatus status = new ProgressStatus(string.Empty);
using (var reader = new StreamReader(fastaPath))
{
proteomeDb.AddFastaFile(reader, new SilentProgressMonitor(), ref status, true); // Delay indexing
}
// perform digestion
proteomeDb.Digest(new ProteaseImpl(trypsin), ProteomeDb.PROTDB_MAX_MISSED_CLEAVAGES, new SilentProgressMonitor(), ref status);
Digestion digestion = proteomeDb.GetDigestion(trypsin.Name);
var digestedProteins0 = digestion.GetProteinsWithSequencePrefix("EDGWVK", 100);
Assert.IsTrue(digestedProteins0.Count >= 1);
}
}
}
public void DoTestOlderProteomeDb(TestContext testContext, bool doActualWebAccess)
{
using (var testFilesDir = new TestFilesDir(testContext, ZIP_FILE))
{
string fastaPath = testFilesDir.GetTestPath("tiny.fasta");
string protDbPath = testFilesDir.GetTestPath("celegans_mini.protdb"); // a version 0 protdb file
string blibPath = testFilesDir.GetTestPath("random.blib"); // a bibliospec file
// What happens when you try to open a random file as a protdb file?
AssertEx.ThrowsException <DbException>(() => ProteomeDb.OpenProteomeDb(fastaPath));
// What happens when you try to open a non-protdb database file as a protdb file?
AssertEx.ThrowsException <FileLoadException>(() => ProteomeDb.OpenProteomeDb(blibPath));
using (ProteomeDb proteomeDb = ProteomeDb.OpenProteomeDb(protDbPath))
{
Assert.IsTrue(proteomeDb.GetSchemaVersionMajor() == 0); // the initial db from our zipfile should be ancient
Assert.IsTrue(proteomeDb.GetSchemaVersionMinor() == 0); // the initial db from our zipfile should be ancient
Assert.AreEqual(9, proteomeDb.GetProteinCount());
var protein = proteomeDb.GetProteinByName("Y18D10A.20");
Assert.IsNotNull(protein);
Assert.IsTrue(String.IsNullOrEmpty(protein.Accession)); // old db won't have this populated
WebEnabledFastaImporter searcher = new WebEnabledFastaImporter(doActualWebAccess ? null :new WebEnabledFastaImporter.FakeWebSearchProvider());
bool searchComplete;
IProgressStatus status = new ProgressStatus(string.Empty);
Assert.IsTrue(proteomeDb.LookupProteinMetadata(new SilentProgressMonitor(), ref status, searcher, false, out searchComplete)); // add any missing protein metadata
Assert.IsTrue(searchComplete);
protein = proteomeDb.GetProteinByName("Y18D10A.20");
Assert.IsNotNull(protein);
if (doActualWebAccess) // We can actually go to the web for metadata
{
Assert.AreEqual("Q9XW16", protein.Accession);
}
using (var reader = new StreamReader(fastaPath))
{
proteomeDb.AddFastaFile(reader, new SilentProgressMonitor(), ref status, false);
}
// the act of writing should update to the current version
Assert.AreEqual(ProteomeDb.SCHEMA_VERSION_MAJOR_CURRENT, proteomeDb.GetSchemaVersionMajor());
Assert.AreEqual(ProteomeDb.SCHEMA_VERSION_MINOR_CURRENT, proteomeDb.GetSchemaVersionMinor());
Assert.AreEqual(19, proteomeDb.GetProteinCount());
// check for propery processed protein metadata
Assert.IsTrue(proteomeDb.LookupProteinMetadata(new SilentProgressMonitor(), ref status, searcher, false, out searchComplete));
Assert.IsTrue(searchComplete);
protein = proteomeDb.GetProteinByName("IPI00000044");
Assert.IsNotNull(protein);
Assert.AreEqual("P01127", protein.Accession); // We get this offline with our ipi->uniprot mapper
if (doActualWebAccess)
{
Assert.AreEqual("PDGFB_HUMAN", protein.PreferredName); // But this we get only with web access
}
/*
* // TODO(bspratt): fix "GetDigestion has no notion of a Db that has been added to, doesn't digest the new proteins and returns immediately (issue #304)"
* Enzyme trypsin = EnzymeList.GetDefault();
* proteomeDb.Digest(trypsin, new SilentProgressMonitor());
* Digestion digestion = proteomeDb.GetDigestion(trypsin.Name);
* var digestedProteins0 = digestion.GetProteinsWithSequencePrefix("EDGWVK", 100);
* Assert.IsTrue(digestedProteins0.Count >= 1);
* */
}
}
}