public List <Modification> get_mods(string current_directory)
{
var psiModDeserialized = Loaders.LoadPsiMod(Path.Combine(current_directory, "Mods", "PSI-MOD.obo.xml"));
Dictionary <string, int> formalChargesDictionary = Loaders.GetFormalChargesDictionary(psiModDeserialized);
Loaders.LoadElements();
List <Modification> all_known_modifications = new List <Modification>();
foreach (string filename in Directory.GetFiles(Path.Combine(current_directory, "Mods")))
{
List <Modification> new_mods = !filename.EndsWith("variable.txt") || Sweet.lollipop.methionine_oxidation ?
PtmListLoader.ReadModsFromFile(filename, formalChargesDictionary, out List <(Modification, string)> filteredModificationsWithWarnings).ToList() :
new List <Modification>(); // Empty variable modifications if not selected
if (filename.EndsWith("variable.txt"))
{
variableModifications = new_mods;
}
if (filename.EndsWith("UniprotGlycanDatabase.txt"))
{
glycan_mods = new_mods;
continue;
}
all_known_modifications.AddRange(new_mods);
}
all_known_modifications = new HashSet <Modification>(all_known_modifications).ToList();
uniprotModifications = make_modification_dictionary(all_known_modifications);
all_mods_with_mass = uniprotModifications.SelectMany(kv => kv.Value).Concat(variableModifications).ToList();
return(all_known_modifications);
}
public static void SetUpModifications()
{
var psiModDeserialized = Loaders.LoadPsiMod(Path.Combine(TestContext.CurrentContext.TestDirectory, "PSI-MOD.obo2.xml"));
Dictionary <string, int> formalChargesDictionary = Loaders.GetFormalChargesDictionary(psiModDeserialized);
UniProtPtms = Loaders.LoadUniprot(Path.Combine(TestContext.CurrentContext.TestDirectory, "ptmlist2.txt"), formalChargesDictionary).ToList();
}
/// <summary>
/// Gets UniProt ptmlist
/// </summary>
/// <param name="spritzDirectory"></param>
/// <returns></returns>
public static List <Modification> GetUniProtMods(string spritzDirectory)
{
Loaders.LoadElements();
var psiModDeserialized = Loaders.LoadPsiMod(Path.Combine(spritzDirectory, "PSI-MOD.obo.xml"));
return(Loaders.LoadUniprot(Path.Combine(spritzDirectory, "ptmlist.txt"), Loaders.GetFormalChargesDictionary(psiModDeserialized)).ToList());
}
private static void LoadModifications()
{
_AllModsKnown = new List <Modification>();
_AllModTypesKnown = new HashSet <string>();
ErrorsReadingMods = new List <string>();
AllModsKnownDictionary = new Dictionary <string, Modification>();
UnimodDeserialized = Loaders.LoadUnimod(Path.Combine(DataDir, @"Data", @"unimod.xml")).ToList();
PsiModDeserialized = Loaders.LoadPsiMod(Path.Combine(DataDir, @"Data", @"PSI-MOD.obo.xml"));
var formalChargesDictionary = Loaders.GetFormalChargesDictionary(PsiModDeserialized);
UniprotDeseralized = Loaders.LoadUniprot(Path.Combine(DataDir, @"Data", @"ptmlist.txt"), formalChargesDictionary).ToList();
foreach (var modFile in Directory.GetFiles(Path.Combine(DataDir, @"Mods")))
{
AddMods(PtmListLoader.ReadModsFromFile(modFile, out var errorMods), false);
}
AddMods(UniprotDeseralized.OfType <Modification>(), false);
AddMods(UnimodDeserialized.OfType <Modification>(), false);
foreach (Modification mod in AllModsKnown)
{
if (!AllModsKnownDictionary.ContainsKey(mod.IdWithMotif))
{
AllModsKnownDictionary.Add(mod.IdWithMotif, mod);
}
// no error thrown if multiple mods with this ID are present - just pick one
}
ProteaseMods = UsefulProteomicsDatabases.PtmListLoader.ReadModsFromFile(Path.Combine(DataDir, @"Mods", @"ProteaseMods.txt"), out var errors).ToList();
ProteaseDictionary.Dictionary = ProteaseDictionary.LoadProteaseDictionary(Path.Combine(DataDir, @"ProteolyticDigestion", @"proteases.tsv"), ProteaseMods);
}
public void FilesLoading()
{
Loaders.LoadElements(Path.Combine(TestContext.CurrentContext.TestDirectory, "elements.dat"));
Loaders.LoadUnimod(Path.Combine(TestContext.CurrentContext.TestDirectory, "unimod_tables.xml"));
Loaders.LoadPsiMod(Path.Combine(TestContext.CurrentContext.TestDirectory, "PSI-MOD.obo.xml"));
Loaders.LoadUniprot(Path.Combine(TestContext.CurrentContext.TestDirectory, "ptmlist.txt"));
}
public void FilesLoading()
{
Loaders.LoadElements(Path.Combine(TestContext.CurrentContext.TestDirectory, "elements2.dat"));
var unimodMods = Loaders.LoadUnimod(Path.Combine(TestContext.CurrentContext.TestDirectory, "unimod_tables2.xml")).ToList();
Assert.AreEqual(2, (unimodMods.First(b => b.id.Equals("HexNAc(2) on N at Any")) as ModificationWithMass).neutralLosses.Count);
var psiModDeserialized = Loaders.LoadPsiMod(Path.Combine(TestContext.CurrentContext.TestDirectory, "PSI-MOD.obo2.xml"));
// N6,N6,N6-trimethyllysine
var trimethylLysine = psiModDeserialized.Items.OfType <UsefulProteomicsDatabases.Generated.oboTerm>().First(b => b.id.Equals("MOD:00083"));
Assert.AreEqual("1+", trimethylLysine.xref_analog.First(b => b.dbname.Equals("FormalCharge")).name);
// Phosphoserine
Assert.IsFalse(psiModDeserialized.Items.OfType <UsefulProteomicsDatabases.Generated.oboTerm>().First(b => b.id.Equals("MOD:00046")).xref_analog.Any(b => b.dbname.Equals("FormalCharge")));
Dictionary <string, int> formalChargesDictionary = Loaders.GetFormalChargesDictionary(psiModDeserialized);
var uniprotPtms = Loaders.LoadUniprot(Path.Combine(TestContext.CurrentContext.TestDirectory, "ptmlist2.txt"), formalChargesDictionary).ToList();
using (StreamWriter w = new StreamWriter(Path.Combine(TestContext.CurrentContext.TestDirectory, "test.txt")))
{
foreach (var nice in uniprotPtms)
{
w.WriteLine(nice.ToString());
w.WriteLine("//");
}
foreach (var nice in unimodMods)
{
w.WriteLine(nice.ToString());
w.WriteLine("//");
}
}
var sampleModList = PtmListLoader.ReadModsFromFile(Path.Combine(TestContext.CurrentContext.TestDirectory, "test.txt")).ToList();
string stringRepresentation = "ID (3R)-3-hydroxyarginine\r\nMT UniProt\r\nPP Anywhere.\r\nDR RESID; AA0601\r\nDR PSI-MOD; MOD:01956\r\nDR UniProt; PTM-0476\r\nTG R\r\nMM 15.994915\r\nCF O";
Assert.AreEqual(stringRepresentation, sampleModList.First().ToString());
// N,N,N-trimethylalanine
var thisMod = sampleModList.First(b => b.id.Equals(@"N,N,N-trimethylalanine")) as ModificationWithMass;
Assert.IsTrue(thisMod.monoisotopicMass > 42);
Assert.IsTrue(thisMod.monoisotopicMass < 43);
}
public void Test_readUniProtXML_writeProteinXml()
{
ModificationMotif.TryGetMotif("X", out ModificationMotif motif);
var nice = new List <Modification>
{
new Modification("fayk", null, "mt", null, motif, "Anywhere.", null, 10, null, null, null, null, null, null)
};
var psiModDeserialized = Loaders.LoadPsiMod(Path.Combine(TestContext.CurrentContext.TestDirectory, "PSI-MOD.obo2.xml"));
Dictionary <string, int> formalChargesDictionary = Loaders.GetFormalChargesDictionary(psiModDeserialized);
var uniprotPtms = Loaders.LoadUniprot(Path.Combine(TestContext.CurrentContext.TestDirectory, "ptmlist2.txt"), formalChargesDictionary).ToList();
List <Protein> ok = ProteinDbLoader.LoadProteinXML(Path.Combine(TestContext.CurrentContext.TestDirectory, "DatabaseTests", @"xml2.xml"), true, DecoyType.None, uniprotPtms.Concat(nice), false, null,
out Dictionary <string, Modification> un);
Protein zero = ok[0];
Protein one = ok[1];
Dictionary <int, List <Modification> > zero_mods = zero.OneBasedPossibleLocalizedModifications as Dictionary <int, List <Modification> >;
Dictionary <int, List <Modification> > one_mods = one.OneBasedPossibleLocalizedModifications as Dictionary <int, List <Modification> >;
ProteinDbWriter.WriteXmlDatabase(new Dictionary <string, HashSet <Tuple <int, Modification> > >(), ok, Path.Combine(TestContext.CurrentContext.TestDirectory, "DatabaseTests", @"rewrite_xml2.xml"));
List <Protein> ok2 = ProteinDbLoader.LoadProteinXML(Path.Combine(TestContext.CurrentContext.TestDirectory, "DatabaseTests", @"rewrite_xml2.xml"), true, DecoyType.None, nice, false,
new List <string>(), out un);
Assert.AreEqual(ok.Count, ok2.Count);
Assert.True(Enumerable.Range(0, ok.Count).All(i => ok[i].BaseSequence == ok2[i].BaseSequence));
Assert.AreEqual(9, ok[0].DatabaseReferences.Count(dbRef => dbRef.Type == "GO"));
Assert.AreEqual(1, ok[0].DatabaseReferences.Count(dbRef => dbRef.Type == "GeneID"));
Assert.AreEqual(3, ok[0].DatabaseReferences.First(dbRef => dbRef.Type == "GO").Properties.Count());
Assert.AreEqual(3, ok[0].GeneNames.Count());
Assert.AreEqual("primary", ok[0].GeneNames.First().Item1);
Assert.AreEqual("JJJ1", ok[0].GeneNames.First().Item2);
Assert.AreEqual("Saccharomyces cerevisiae (strain ATCC 204508 / S288c)", ok[0].Organism);
Assert.AreEqual(Path.Combine(TestContext.CurrentContext.TestDirectory, "DatabaseTests", @"xml2.xml"), ok[0].DatabaseFilePath);
Assert.AreEqual(9, ok2[0].DatabaseReferences.Count(dbRef => dbRef.Type == "GO"));
Assert.AreEqual(3, ok2[0].DatabaseReferences.First(dbRef => dbRef.Type == "GO").Properties.Count());
Assert.AreEqual(3, ok2[0].GeneNames.Count());
Assert.AreEqual("primary", ok2[0].GeneNames.First().Item1);
Assert.AreEqual("JJJ1", ok2[0].GeneNames.First().Item2);
Assert.AreEqual("Saccharomyces cerevisiae (strain ATCC 204508 / S288c)", ok2[0].Organism);
Assert.AreEqual(Path.Combine(TestContext.CurrentContext.TestDirectory, "DatabaseTests", @"rewrite_xml2.xml"), ok2[0].DatabaseFilePath);
Assert.True(ok.All(p => p.ProteolysisProducts.All(prod => prod.OneBasedBeginPosition == null || prod.OneBasedBeginPosition > 0 && prod.OneBasedBeginPosition <= p.Length)));
Assert.True(ok.All(p => p.ProteolysisProducts.All(prod => prod.OneBasedEndPosition == null || prod.OneBasedEndPosition > 0 && prod.OneBasedEndPosition <= p.Length)));
Assert.True(ok2.All(p => p.ProteolysisProducts.All(prod => prod.OneBasedBeginPosition == null || prod.OneBasedBeginPosition > 0 && prod.OneBasedBeginPosition <= p.Length)));
Assert.True(ok2.All(p => p.ProteolysisProducts.All(prod => prod.OneBasedEndPosition == null || prod.OneBasedEndPosition > 0 && prod.OneBasedEndPosition <= p.Length)));
}
public void Test_write_with_custom_mods()
{
ModificationMotif.TryGetMotif("S", out ModificationMotif m1);
ModificationMotif.TryGetMotif("T", out ModificationMotif m2);
ModificationMotif.TryGetMotif("X", out ModificationMotif motiff);
var nice = new List <Modification>
{
new Modification("fayk", null, "mt", null, motiff, "Anywhere.", null, 10, null, null, null, null, null, null),
new Modification("Phosphoserine", null, "mt", null, m1, "Anywhere.", null, 80, null, null, null, null, null, null),
new Modification("Phosphothreonine", null, "mt", null, m2, "Anywhere.", null, 80, null, null, null, null, null, null)
};
ModificationMotif.TryGetMotif("K", out ModificationMotif motif);
Modification m = new Modification("mod", null, "mt", null, motif, "Anywhere.", null, 1, null, null, null, new Dictionary <DissociationType, List <double> >()
{
{ DissociationType.AnyActivationType, new List <double> {
-1
} }
}, null, null);
Dictionary <string, HashSet <Tuple <int, Modification> > > new_mods = new Dictionary <string, HashSet <Tuple <int, Modification> > >
{
{ "P53863", new HashSet <Tuple <int, Modification> > {
new Tuple <int, Modification>(2, m)
} }
};
var psiModDeserialized = Loaders.LoadPsiMod(Path.Combine(TestContext.CurrentContext.TestDirectory, "PSI-MOD.obo2.xml"));
Dictionary <string, int> formalChargesDictionary = Loaders.GetFormalChargesDictionary(psiModDeserialized);
var uniprotPtms = Loaders.LoadUniprot(Path.Combine(TestContext.CurrentContext.TestDirectory, "ptmlist2.txt"), formalChargesDictionary).ToList();
List <Protein> ok = ProteinDbLoader.LoadProteinXML(Path.Combine(TestContext.CurrentContext.TestDirectory, "DatabaseTests", @"xml2.xml"), true, DecoyType.None, uniprotPtms.Concat(nice), false, new List <string>(),
out Dictionary <string, Modification> un);
var newModResEntries = ProteinDbWriter.WriteXmlDatabase(new_mods, ok, Path.Combine(TestContext.CurrentContext.TestDirectory, "DatabaseTests", @"rewrite_xml2.xml"));
Assert.AreEqual(1, newModResEntries.Count);
List <Protein> ok2 = ProteinDbLoader.LoadProteinXML(Path.Combine(TestContext.CurrentContext.TestDirectory, "DatabaseTests", @"rewrite_xml2.xml"), true, DecoyType.None,
nice, false, new List <string>(), out un);
Assert.AreEqual(ok.Count, ok2.Count);
Assert.True(Enumerable.Range(0, ok.Count).All(i => ok[i].BaseSequence == ok2[i].BaseSequence));
Assert.AreEqual(2, ok[0].OneBasedPossibleLocalizedModifications.Count);
Assert.AreEqual(3, ok2[0].OneBasedPossibleLocalizedModifications.Count);
}
public void FilesLoading() //delete mzLib\Test\bin\x64\Debug to update your local unimod list
{
Loaders.LoadElements();
string uniModPath = Path.Combine(TestContext.CurrentContext.TestDirectory, "unimod_tables2.xml");
string psiModPath = Path.Combine(TestContext.CurrentContext.TestDirectory, "PSI-MOD.obo2.xml");
string uniProtPath = Path.Combine(TestContext.CurrentContext.TestDirectory, "ptmlist2.txt");
// UniModPTMs
var unimodMods = Loaders.LoadUnimod(uniModPath).ToList();
Assert.AreEqual(2677, unimodMods.Count); // UniMod PTM list may be updated at some point, causing the unit test to fail
List <Modification> myList = unimodMods.Where(m => m.OriginalId.Equals("HexNAc(2)")).ToList();
Modification testMod = myList.First();
int neutralLossCount = 0;
if (testMod.NeutralLosses.Count != 0)
{
foreach (KeyValuePair <MassSpectrometry.DissociationType, List <double> > item in testMod.NeutralLosses)
{
foreach (double loss in item.Value)
{
neutralLossCount++;
}
}
}
Assert.AreEqual(2, neutralLossCount);
// PsiMod PTMs
var psiModDeserialized = Loaders.LoadPsiMod(psiModPath);
// N6,N6,N6-trimethyllysine
var trimethylLysine = psiModDeserialized.Items.OfType <UsefulProteomicsDatabases.Generated.oboTerm>().First(b => b.id.Equals("MOD:00083"));
Assert.AreEqual("1+", trimethylLysine.xref_analog.First(b => b.dbname.Equals("FormalCharge")).name);
// Phosphoserine
Assert.IsFalse(psiModDeserialized.Items.OfType <UsefulProteomicsDatabases.Generated.oboTerm>().First(b => b.id.Equals("MOD:00046")).xref_analog.Any(b => b.dbname.Equals("FormalCharge")));
Dictionary <string, int> formalChargesDictionary = Loaders.GetFormalChargesDictionary(psiModDeserialized);
// UniProt PTMs
var uniprotPtms = Loaders.LoadUniprot(uniProtPath, formalChargesDictionary).ToList();
Assert.AreEqual(345, uniprotPtms.Count()); // UniProt PTM list may be updated at some point, causing the unit test to fail
// write UniProt and UniMod PTMs to a file
using (StreamWriter w = new StreamWriter(Path.Combine(TestContext.CurrentContext.TestDirectory, "test.txt")))
{
foreach (var nice in uniprotPtms)
{
w.WriteLine(nice.ToString());
w.WriteLine("//");
}
foreach (var nice in unimodMods)
{
w.WriteLine(nice.ToString());
w.WriteLine("//");
}
}
// read in the file and make sure that it has the same number of PTMs
var sampleModList = PtmListLoader.ReadModsFromFile(Path.Combine(TestContext.CurrentContext.TestDirectory, "test.txt"), out var errors).ToList();
Assert.AreEqual(uniprotPtms.Count + unimodMods.Count, sampleModList.Count());
List <Modification> myOtherList = new List <Modification>();
foreach (Modification mod in sampleModList)
{
if (mod.IdWithMotif != null && mod.IdWithMotif.Contains("Acetyl"))
{
myOtherList.Add(mod);
}
}
var thisMod = myOtherList.First();
Assert.IsTrue(thisMod.MonoisotopicMass > 42);
Assert.IsTrue(thisMod.MonoisotopicMass < 43);
File.Delete(uniModPath);
File.Delete(psiModPath);
File.Delete(uniProtPath);
}
public void get_theoretical_proteoforms(string current_directory)
{
if (!ready_to_make_database(current_directory))
{
return;
}
//Clear out data from potential previous runs
foreach (ProteoformCommunity community in Sweet.lollipop.decoy_proteoform_communities.Values)
{
community.theoretical_proteoforms = new TheoreticalProteoform[0];
}
theoretical_proteins.Clear();
//Read the UniProt-XML and ptmlist
var psiModDeserialized = Loaders.LoadPsiMod(Path.Combine(current_directory, "Mods", "PSI-MOD.obo.xml"));
Dictionary <string, int> formalChargesDictionary = Loaders.GetFormalChargesDictionary(psiModDeserialized);
List <Modification> all_known_modifications = Sweet.lollipop.get_files(Sweet.lollipop.input_files, Purpose.PtmList)
.SelectMany(file => PtmListLoader.ReadModsFromFile(file.complete_path, formalChargesDictionary, out List <(Modification, string)> filteredModificationsWithWarnings))
.ToList();
uniprotModifications = make_modification_dictionary(all_known_modifications);
Parallel.ForEach(Sweet.lollipop.get_files(Sweet.lollipop.input_files, Purpose.ProteinDatabase).ToList(), database =>
{
if (database.extension == ".xml")
{
lock (theoretical_proteins)
theoretical_proteins.Add(database, ProteinDbLoader.LoadProteinXML(database.complete_path, true, DecoyType.None, all_known_modifications, database.ContaminantDB, Sweet.lollipop.mod_types_to_exclude, out Dictionary <string, Modification> um).ToArray());
lock (all_known_modifications) all_known_modifications.AddRange(ProteinDbLoader.GetPtmListFromProteinXml(database.complete_path).Where(m => !Sweet.lollipop.mod_types_to_exclude.Contains(m.ModificationType)));
}
else if (database.extension == ".fasta")
{
lock (theoretical_proteins)
theoretical_proteins.Add(database, ProteinDbLoader.LoadProteinFasta(database.complete_path, true, DecoyType.None, database.ContaminantDB, ProteinDbLoader.UniprotAccessionRegex, ProteinDbLoader.UniprotFullNameRegex, ProteinDbLoader.UniprotFullNameRegex, ProteinDbLoader.UniprotGeneNameRegex,
ProteinDbLoader.UniprotOrganismRegex, out var dbErrors).ToArray());
}
});
foreach (string filename in Directory.GetFiles(Path.Combine(current_directory, "Mods")))
{
List <Modification> new_mods = !filename.EndsWith("variable.txt") || Sweet.lollipop.methionine_oxidation ?
PtmListLoader.ReadModsFromFile(filename, formalChargesDictionary, out List <(Modification, string)> filteredModificationsWithWarnings).ToList() :
new List <Modification>(); // Empty variable modifications if not selected
if (filename.EndsWith("variable.txt"))
{
variableModifications = new_mods;
}
all_known_modifications.AddRange(new_mods);
}
all_known_modifications = new HashSet <Modification>(all_known_modifications).ToList();
uniprotModifications = make_modification_dictionary(all_known_modifications);
all_mods_with_mass = uniprotModifications.SelectMany(kv => kv.Value).Concat(variableModifications).ToList();
Sweet.lollipop.modification_ranks = rank_mods(theoretical_proteins, variableModifications, all_mods_with_mass);
unlocalized_lookup = make_unlocalized_lookup(all_mods_with_mass.Concat(new List <Modification> {
new Ptm().modification
}));
load_unlocalized_names(Path.Combine(Environment.CurrentDirectory, "Mods", "stored_mods.modnames"));
//this is for ptmsets --> used in RELATIONS
all_possible_ptmsets = PtmCombos.generate_all_ptmsets(2, all_mods_with_mass, Sweet.lollipop.modification_ranks, Sweet.lollipop.mod_rank_first_quartile / 2).ToList();
for (int i = 2; i <= Math.Max(ptmset_max_number_of_a_kind, Sweet.lollipop.max_ptms); i++) // the method above doesn't make 2 or more of a kind, so we make it here
{
all_possible_ptmsets.AddRange(all_mods_with_mass.Select(m => new PtmSet(Enumerable.Repeat(new Ptm(-1, m), i).ToList(), Sweet.lollipop.modification_ranks, Sweet.lollipop.mod_rank_first_quartile / 2)));
}
//Generate lookup table for ptm sets based on rounded mass of eligible PTMs -- used in forming ET relations
possible_ptmset_dictionary = make_ptmset_dictionary();
make_theoretical_proteoforms();
}