private void QueryPeptideProteins(ILongWaitBroker longWaitBroker)
{
List <HashSet <Protein> > peptideProteins = new List <HashSet <Protein> >();
if (BackgroundProteome != null)
{
using (var proteomeDb = BackgroundProteome.OpenProteomeDb(longWaitBroker.CancellationToken))
{
Digestion digestion = proteomeDb.GetDigestion();
if (digestion != null)
{
var peptidesOfInterest = _peptideDocNodes.Select(node => node.Item2.Peptide.Target.Sequence);
var sequenceProteinsDict = digestion.GetProteinsWithSequences(peptidesOfInterest);
if (longWaitBroker.IsCanceled)
{
return;
}
foreach (var tuple in _peptideDocNodes)
{
if (longWaitBroker.IsCanceled)
{
return;
}
var peptideGroup = (PeptideGroup)tuple.Item1.GetIdentity(0);
var peptideDocNode = tuple.Item2;
HashSet <Protein> proteins = new HashSet <Protein>();
var peptideGroupDocNode = PeptideGroupDocNodes.First(g => ReferenceEquals(g.PeptideGroup, peptideGroup));
List <Protein> proteinsForSequence;
if (sequenceProteinsDict.TryGetValue(peptideDocNode.Peptide.Target.Sequence, out proteinsForSequence))
{
if (peptideGroupDocNode != null)
{
foreach (var protein in proteinsForSequence)
{
if (protein.Sequence == peptideGroupDocNode.PeptideGroup.Sequence)
{
_peptidesInBackgroundProteome.Add(tuple.Item1);
continue;
}
proteins.Add(protein);
}
}
}
peptideProteins.Add(proteins);
}
}
}
}
_peptideProteins = peptideProteins;
}
void dataGridView1_CurrentCellChanged(object sender, EventArgs e)
{
if (dataGridView1.CurrentCell == null || dataGridView1.CurrentRow == null)
{
return;
}
var rowTag = (Tuple <IdentityPath, PeptideDocNode>)dataGridView1.CurrentRow.Tag;
if (rowTag == null)
{
return;
}
PeptideDocNode peptideDocNode = rowTag.Item2;
// Expecting to find this peptide
var peptideGroupDocNode = PeptideGroupDocNodes.First(g => g.Peptides.Contains(peptideDocNode));
String peptideSequence = peptideDocNode.Peptide.Target.Sequence;
String proteinSequence;
var proteinColumn = dataGridView1.Columns[dataGridView1.CurrentCell.ColumnIndex].Tag as ProteinColumn;
ProteinMetadata metadata;
if (proteinColumn == null)
{
metadata = peptideGroupDocNode.ProteinMetadata;
proteinSequence = peptideGroupDocNode.PeptideGroup.Sequence;
}
else
{
metadata = proteinColumn.Protein.ProteinMetadata;
proteinSequence = proteinColumn.Protein.Sequence;
}
tbxProteinName.Text = metadata.Name;
tbxProteinDescription.Text = metadata.Description;
tbxProteinDetails.Text = metadata.DisplayTextWithoutNameOrDescription(); // Don't show name or description
if (!string.IsNullOrEmpty(proteinSequence))
{
var regex = new Regex(peptideSequence);
StringBuilder formattedText = new StringBuilder("{\\rtf1\\ansi{\\fonttbl\\f0\\fswiss Helvetica;}{\\colortbl ;\\red0\\green0\\blue255;}\\f0\\pard \\fs16"); // Not L10N
int lastIndex = 0;
for (Match match = regex.Match(proteinSequence, 0); match.Success; lastIndex = match.Index + match.Length, match = match.NextMatch())
{
formattedText.Append("\\cf0\\b0 " + proteinSequence.Substring(lastIndex, match.Index - lastIndex)); // Not L10N
formattedText.Append("\\cf1\\b " + proteinSequence.Substring(match.Index, match.Length)); // Not L10N
}
formattedText.Append("\\cf0\\b0 " + proteinSequence.Substring(lastIndex, proteinSequence.Length - lastIndex)); // Not L10N
formattedText.Append("\\par }"); // Not L10N
richTextBoxSequence.Rtf = formattedText.ToString();
}
}
private void QueryPeptideProteins(ILongWaitBroker longWaitBroker)
{
List <HashSet <Protein> > peptideProteins = new List <HashSet <Protein> >();
if (BackgroundProteome != null)
{
using (var proteomeDb = BackgroundProteome.OpenProteomeDb())
{
Digestion digestion = BackgroundProteome.GetDigestion(proteomeDb, SrmDocument.Settings.PeptideSettings);
if (digestion != null)
{
var peptidesOfInterest = _peptideDocNodes.Select(node => node.Peptide.Sequence);
var sequenceProteinsDict = digestion.GetProteinsWithSequences(peptidesOfInterest, null);
if (longWaitBroker.IsCanceled)
{
return;
}
foreach (var peptideDocNode in _peptideDocNodes)
{
HashSet <Protein> proteins = new HashSet <Protein>();
var peptideGroupDocNode = PeptideGroupDocNodes.First(g => g.Peptides.Contains(peptideDocNode));
List <Protein> proteinsForSequence;
if (sequenceProteinsDict.TryGetValue(peptideDocNode.Peptide.Sequence, out proteinsForSequence))
{
if (peptideGroupDocNode != null)
{
foreach (var protein in proteinsForSequence)
{
if (protein.Sequence == peptideGroupDocNode.PeptideGroup.Sequence)
{
_peptidesInBackgroundProteome.Add(peptideDocNode);
continue;
}
proteins.Add(protein);
}
}
}
peptideProteins.Add(proteins);
}
}
}
}
_peptideProteins = peptideProteins;
}
private void SelectPeptidesWithNumberOfMatchesAtOrBelowThreshold(int threshold, UniquenessType uniqueBy)
{
dataGridView1.EndEdit();
var dubious = new HashSet <string>(StringComparer.OrdinalIgnoreCase);
for (int rowIndex = 0; rowIndex < dataGridView1.Rows.Count; rowIndex++)
{
var row = dataGridView1.Rows[rowIndex];
var rowTag = (Tuple <IdentityPath, PeptideDocNode>)row.Tag;
int matchCount = _peptidesInBackgroundProteome.Contains(rowTag.Item1) ? 1 : 0;
for (int col = 0; col < dataGridView1.ColumnCount; col++)
{
if (col == PeptideIncludedColumn.Index || col == PeptideColumn.Index)
{
continue;
}
if (row.Cells[col].Value is bool && ((bool)row.Cells[col].Value))
{
if (uniqueBy == UniquenessType.protein)
{
matchCount++;
}
else
{
var peptide = rowTag.Item2;
var parent = PeptideGroupDocNodes.First(p => p.Children.Contains(peptide));
string testValA;
string testValB;
// ATP5B and atp5b are the same thing, as are "mus musculus" and "MUS MUSCULUS"
if (uniqueBy == UniquenessType.gene)
{
// ReSharper disable once PossibleNullReferenceException
testValA = parent.ProteinMetadata.Gene;
testValB = ((ProteinColumn)dataGridView1.Columns[col].Tag).Protein.Gene;
}
else
{
// ReSharper disable once PossibleNullReferenceException
testValA = parent.ProteinMetadata.Species;
testValB = ((ProteinColumn)dataGridView1.Columns[col].Tag).Protein.Species;
}
// Can't filter on something that isn't there - require nonempty values
if (!string.IsNullOrEmpty(testValA) && !string.IsNullOrEmpty(testValB) &&
string.Compare(testValA, testValB, StringComparison.OrdinalIgnoreCase) != 0)
{
matchCount++;
}
if (string.IsNullOrEmpty(testValA))
{
dubious.Add(parent.Name);
}
if (string.IsNullOrEmpty(testValB))
{
dubious.Add(((ProteinColumn)dataGridView1.Columns[col].Tag).Protein.Name);
}
}
}
if (matchCount > threshold)
{
break;
}
}
row.Cells[PeptideIncludedColumn.Name].Value = (matchCount <= threshold);
}
SetCheckBoxPeptideIncludedHeaderState();
if (dubious.Any())
{
var dubiousValues = TextUtil.LineSeparate(uniqueBy == UniquenessType.gene ?
Resources.UniquePeptidesDlg_SelectPeptidesWithNumberOfMatchesAtOrBelowThreshold_Some_background_proteome_proteins_did_not_have_gene_information__this_selection_may_be_suspect_ :
Resources.UniquePeptidesDlg_SelectPeptidesWithNumberOfMatchesAtOrBelowThreshold_Some_background_proteome_proteins_did_not_have_species_information__this_selection_may_be_suspect_,
Resources.UniquePeptidesDlg_SelectPeptidesWithNumberOfMatchesAtOrBelowThreshold_These_proteins_include_,
TextUtil.LineSeparate(dubious)); // Not L10N
MessageDlg.Show(this, dubiousValues);
}
}
