public void test_construct_target_and_decoy_families()
{
SaveState.lollipop = new Lollipop();
SaveState.lollipop.neucode_labeled = true;
ExperimentalProteoform pf1 = ConstructorsForTesting.ExperimentalProteoform("E1", 1000, 5, true);
ExperimentalProteoform pf2 = ConstructorsForTesting.ExperimentalProteoform("E2", 1020, 5, true);
ExperimentalProteoform pf3 = ConstructorsForTesting.ExperimentalProteoform("E3", 1040, 8, true);
ExperimentalProteoform pf4 = ConstructorsForTesting.ExperimentalProteoform("E4", 1050, 5, true);
TheoreticalProteoform t1 = ConstructorsForTesting.make_a_theoretical("t1", 1000, 5);
TheoreticalProteoform decoy1 = ConstructorsForTesting.make_a_theoretical("Decoy1", 1020, 5);
TestProteoformCommunityRelate.prepare_for_et(new List <double>()
{
0
});
SaveState.lollipop.target_proteoform_community = new ProteoformCommunity();
SaveState.lollipop.target_proteoform_community.experimental_proteoforms = new ExperimentalProteoform[4] {
pf1, pf2, pf3, pf4
};
SaveState.lollipop.target_proteoform_community.theoretical_proteoforms = new TheoreticalProteoform[1] {
t1
};
SaveState.lollipop.decoy_proteoform_communities.Add(SaveState.lollipop.decoy_community_name_prefix + "0", new ProteoformCommunity());
SaveState.lollipop.decoy_proteoform_communities[SaveState.lollipop.decoy_community_name_prefix + "0"].experimental_proteoforms = SaveState.lollipop.target_proteoform_community.experimental_proteoforms.Select(e => new ExperimentalProteoform(e)).ToArray();
SaveState.lollipop.decoy_proteoform_communities[SaveState.lollipop.decoy_community_name_prefix + "0"].theoretical_proteoforms = new TheoreticalProteoform[1] {
decoy1
};
SaveState.lollipop.et_relations = SaveState.lollipop.target_proteoform_community.relate(SaveState.lollipop.target_proteoform_community.experimental_proteoforms, SaveState.lollipop.target_proteoform_community.theoretical_proteoforms, ProteoformComparison.ExperimentalTheoretical, true, TestContext.CurrentContext.TestDirectory, true);
SaveState.lollipop.ee_relations = SaveState.lollipop.target_proteoform_community.relate(SaveState.lollipop.target_proteoform_community.experimental_proteoforms, SaveState.lollipop.target_proteoform_community.experimental_proteoforms, ProteoformComparison.ExperimentalExperimental, true, TestContext.CurrentContext.TestDirectory, true);
SaveState.lollipop.relate_ed();
SaveState.lollipop.relate_ef();
foreach (ProteoformRelation pr in SaveState.lollipop.et_relations)
{
pr.set_nearby_group(SaveState.lollipop.et_relations, SaveState.lollipop.et_relations.Select(r => r.InstanceId).ToList());
}
foreach (ProteoformRelation pr in SaveState.lollipop.ee_relations)
{
pr.set_nearby_group(SaveState.lollipop.ee_relations, SaveState.lollipop.ee_relations.Select(r => r.InstanceId).ToList());
}
SaveState.lollipop.et_peaks = SaveState.lollipop.target_proteoform_community.accept_deltaMass_peaks(SaveState.lollipop.et_relations, SaveState.lollipop.ed_relations);
SaveState.lollipop.ee_peaks = SaveState.lollipop.target_proteoform_community.accept_deltaMass_peaks(SaveState.lollipop.ee_relations, SaveState.lollipop.ef_relations);
//one ED relation
Assert.AreEqual(1, SaveState.lollipop.ed_relations.Values.SelectMany(v => v).Count());
Assert.AreEqual(1, SaveState.lollipop.decoy_proteoform_communities[SaveState.lollipop.decoy_community_name_prefix + "0"].theoretical_proteoforms.SelectMany(t => t.relationships).Count());
//peak is unaccepted --> relation should be unaccepted, peak added to relation
Assert.IsFalse(SaveState.lollipop.ed_relations.First().Value.FirstOrDefault().Accepted); //should be false if peak unaccepted
Assert.AreEqual(1, SaveState.lollipop.et_peaks.Count);
Assert.IsNotNull(SaveState.lollipop.ed_relations.Values.SelectMany(v => v).First().peak);
//peak is accepted --> relation should be accepted, peak added to relation
SaveState.lollipop.clear_et();
SaveState.lollipop.min_peak_count_et = 1;
SaveState.lollipop.et_relations = SaveState.lollipop.target_proteoform_community.relate(SaveState.lollipop.target_proteoform_community.experimental_proteoforms, SaveState.lollipop.target_proteoform_community.theoretical_proteoforms, ProteoformComparison.ExperimentalTheoretical, true, TestContext.CurrentContext.TestDirectory, true);
SaveState.lollipop.relate_ed();
foreach (ProteoformRelation pr in SaveState.lollipop.et_relations)
{
pr.set_nearby_group(SaveState.lollipop.et_relations, SaveState.lollipop.et_relations.Select(r => r.InstanceId).ToList());
}
foreach (ProteoformRelation pr in SaveState.lollipop.ee_relations)
{
pr.set_nearby_group(SaveState.lollipop.ee_relations, SaveState.lollipop.ee_relations.Select(r => r.InstanceId).ToList());
}
SaveState.lollipop.et_peaks = SaveState.lollipop.target_proteoform_community.accept_deltaMass_peaks(SaveState.lollipop.et_relations, SaveState.lollipop.ed_relations);
SaveState.lollipop.et_peaks = SaveState.lollipop.target_proteoform_community.accept_deltaMass_peaks(SaveState.lollipop.et_relations, SaveState.lollipop.ed_relations);
Assert.IsTrue(SaveState.lollipop.ed_relations.First().Value.FirstOrDefault().Accepted); //should be true if peak accepted
Assert.IsNotNull(SaveState.lollipop.ed_relations.Values.SelectMany(v => v).First().peak);
Assert.AreEqual(1, SaveState.lollipop.ed_relations.Values.SelectMany(v => v).Count(r => r.Accepted)); //only 1 relation is in delta mass range of accepted peak - only 1 accepted
//2 ef relations, unaccepted because peak unaccepted
Assert.AreEqual(3, SaveState.lollipop.ef_relations.Values.SelectMany(v => v).Count()); //2 ef relations
Assert.AreEqual(0, SaveState.lollipop.ef_relations.Values.SelectMany(v => v).Count(r => r.Accepted)); //all peaks accepted are false, so relation accepted should be false
//one peak accepted --> one of the EF relations falls into range of peak and should be accepted
SaveState.lollipop.clear_ee();
SaveState.lollipop.min_peak_count_ee = 1;
SaveState.lollipop.ee_relations = SaveState.lollipop.target_proteoform_community.relate(SaveState.lollipop.target_proteoform_community.experimental_proteoforms, SaveState.lollipop.target_proteoform_community.experimental_proteoforms, ProteoformComparison.ExperimentalExperimental, true, TestContext.CurrentContext.TestDirectory, true);
SaveState.lollipop.relate_ef();
SaveState.lollipop.ee_peaks = SaveState.lollipop.target_proteoform_community.accept_deltaMass_peaks(SaveState.lollipop.ee_relations, SaveState.lollipop.ef_relations);
Assert.AreEqual(3, SaveState.lollipop.ee_peaks.Count(p => p.Accepted));
Assert.AreEqual(3, SaveState.lollipop.ee_relations.Count);
Assert.AreEqual(1, SaveState.lollipop.ef_relations.Values.SelectMany(v => v).Count(r => r.Accepted)); //only 1 relation is in delta mass range of accepted peak - only 1 accepted
SaveState.lollipop.construct_target_and_decoy_families();
//should be 2 target families and 3 decoy families
//only make decoy relations out of accepted relations --> 1 family with accepted relations
Assert.AreEqual(2, SaveState.lollipop.target_proteoform_community.families.Count);
Assert.AreEqual(3, SaveState.lollipop.decoy_proteoform_communities[SaveState.lollipop.decoy_community_name_prefix + "0"].families.Count);
Assert.AreEqual(1, SaveState.lollipop.decoy_proteoform_communities[SaveState.lollipop.decoy_community_name_prefix + "0"].families.Count(f => f.relations.Count > 0));
Assert.AreEqual(2, SaveState.lollipop.decoy_proteoform_communities[SaveState.lollipop.decoy_community_name_prefix + "0"].families.SelectMany(f => f.relations).Count());
}
public void TestDeltaMassPeakConstructorWithNotches()
{
Sweet.lollipop = new Lollipop();
Sweet.lollipop.et_use_notch = true;
Sweet.lollipop.enter_input_files(new string[] { Path.Combine(TestContext.CurrentContext.TestDirectory, "ptmlist.txt") }, Lollipop.acceptable_extensions[2], Lollipop.file_types[2], Sweet.lollipop.input_files, false);
ConstructorsForTesting.read_mods();
Sweet.lollipop.et_high_mass_difference = 250;
Sweet.lollipop.et_low_mass_difference = -250;
ExperimentalProteoform pf1 = ConstructorsForTesting.ExperimentalProteoform("acession1");
TheoreticalProteoform pf2 = ConstructorsForTesting.make_a_theoretical();
pf1.modified_mass = 1000;
ProteoformComparison relation_type = ProteoformComparison.ExperimentalTheoretical;
double delta_mass = 1 - 1e-7;
ExperimentalProteoform pf3 = ConstructorsForTesting.ExperimentalProteoform("acession3");
TheoreticalProteoform pf4 = ConstructorsForTesting.make_a_theoretical();
pf3.modified_mass = 1000;
ProteoformComparison relation_type2 = ProteoformComparison.ExperimentalTheoretical;
double delta_mass2 = 1;
ExperimentalProteoform pf5 = ConstructorsForTesting.ExperimentalProteoform("acession5");
TheoreticalProteoform pf6 = ConstructorsForTesting.make_a_theoretical();
pf5.modified_mass = 1000;
ProteoformComparison relation_type3 = ProteoformComparison.ExperimentalTheoretical;
double delta_mass3 = 1 + 1e-7;
ExperimentalProteoform pf55 = ConstructorsForTesting.ExperimentalProteoform("acession5");
TheoreticalProteoform pf65 = ConstructorsForTesting.make_a_theoretical();
pf55.modified_mass = 1000;
ProteoformComparison relation_type35 = ProteoformComparison.ExperimentalTheoretical;
double delta_mass35 = 1 + 2e-7;
TestProteoformCommunityRelate.prepare_for_et(new List <double>()
{
1
});
List <ProteoformRelation> theList = new List <ProteoformRelation>();
theList.Add(new ProteoformRelation(pf1, pf2, relation_type, delta_mass, TestContext.CurrentContext.TestDirectory));
theList.Add(new ProteoformRelation(pf3, pf4, relation_type2, delta_mass2, TestContext.CurrentContext.TestDirectory));
theList.Add(new ProteoformRelation(pf5, pf6, relation_type3, delta_mass3, TestContext.CurrentContext.TestDirectory));
theList.Add(new ProteoformRelation(pf55, pf65, relation_type35, delta_mass35, TestContext.CurrentContext.TestDirectory));
ProteoformRelation base_relation = new ProteoformRelation(pf3, pf4, relation_type2, delta_mass2, TestContext.CurrentContext.TestDirectory);
//base_relation.nearby_relations = base_relation.set_nearby_group(theList, theList.Select(r => r.InstanceId).ToList());
Console.WriteLine("Creating deltaMassPeak");
DeltaMassPeak deltaMassPeak = new DeltaMassPeak(base_relation, new HashSet <ProteoformRelation>(theList));
Console.WriteLine("Created deltaMassPeak");
Assert.AreEqual(0, deltaMassPeak.peak_group_fdr);
Assert.AreEqual(4, deltaMassPeak.grouped_relations.Count);
Assert.AreEqual("[fake1]", deltaMassPeak.possiblePeakAssignments_string);
Assert.AreEqual(1.0, deltaMassPeak.DeltaMass);
Dictionary <string, List <ProteoformRelation> > decoy_relations = new Dictionary <string, List <ProteoformRelation> >();
decoy_relations["decoyDatabase1"] = new List <ProteoformRelation>();
ExperimentalProteoform pf7 = ConstructorsForTesting.ExperimentalProteoform("experimental1");
TheoreticalProteoform pf8 = ConstructorsForTesting.make_a_theoretical();
pf7.modified_mass = 1000;
ProteoformComparison relation_type4 = ProteoformComparison.ExperimentalDecoy;
double delta_mass4 = 1;
ProteoformRelation decoy_relation = new ProteoformRelation(pf7, pf8, relation_type4, delta_mass4, TestContext.CurrentContext.TestDirectory);
decoy_relations["decoyDatabase1"].Add(decoy_relation);
deltaMassPeak.calculate_fdr(decoy_relations);
Assert.AreEqual(0.25, deltaMassPeak.peak_group_fdr); // 1 decoy database, (1 decoy relation, median=1), 4 target relations
decoy_relations["decoyDatabase2"] = new List <ProteoformRelation>();
decoy_relations["decoyDatabase2"].Add(decoy_relation);
decoy_relations["decoyDatabase2"].Add(decoy_relation);
deltaMassPeak.calculate_fdr(decoy_relations);
Assert.AreEqual(0.375, deltaMassPeak.peak_group_fdr); // 2 decoy databases (1 & 2 decoy relations, median=1.5), 4 target relations
}
public void TestRelateTD()
{
Sweet.lollipop = new Lollipop();
Sweet.lollipop.neucode_labeled = false;
Sweet.lollipop.maximum_missed_monos = 1;
Sweet.lollipop.agg_minBiorepsWithObservations = 0;
InputFile f = new InputFile("path", Purpose.Identification);
Sweet.lollipop.input_files.Add(f);
// Two proteoforms; lysine count equal; mass difference < 250 -- return 1
Component c1 = new Component();
c1.weighted_monoisotopic_mass = 1000.0;
c1.rt_apex = 45;
c1.accepted = true;
c1.id = 1.ToString();
c1.intensity_sum = 1e6;
c1.input_file = f;
c1.charge_states = new List <ChargeState>()
{
new ChargeState(1, c1.intensity_sum, c1.weighted_monoisotopic_mass)
};
Component c2 = new Component();
c2.weighted_monoisotopic_mass = 1000.0;
c2.rt_apex = 85;
c2.accepted = true;
c2.input_file = f;
c2.intensity_sum = 1e6;
c2.charge_states = new List <ChargeState>()
{
new ChargeState(1, c2.intensity_sum, c2.weighted_monoisotopic_mass)
};
c2.id = 2.ToString();
Component c3 = new Component();
c3.weighted_monoisotopic_mass = 1131.04;
c3.rt_apex = 45;
c3.accepted = true;
c3.input_file = f;
c3.intensity_sum = 1e6;
c3.charge_states = new List <ChargeState>()
{
new ChargeState(1, c3.intensity_sum, c3.weighted_monoisotopic_mass)
};
c3.id = 3.ToString();
Component c4 = new Component();
c4.weighted_monoisotopic_mass = 2000.00;
c4.rt_apex = 45;
c4.accepted = true;
c4.input_file = new InputFile("path", Purpose.Identification);
c4.intensity_sum = 1e6;
c4.charge_states = new List <ChargeState>()
{
new ChargeState(1, c4.intensity_sum, c4.weighted_monoisotopic_mass)
};
c4.id = 4.ToString();
Component c5 = new Component();
c5.weighted_monoisotopic_mass = 1001.0;
c5.rt_apex = 45;
c5.accepted = true;
c5.input_file = f;
c5.intensity_sum = 1e6;
c5.charge_states = new List <ChargeState>()
{
new ChargeState(1, c5.intensity_sum, c5.weighted_monoisotopic_mass)
};
c5.id = 2.ToString();
List <IAggregatable> components = new List <IAggregatable>()
{
c1, c2, c3, c4, c5
};
Sweet.lollipop.raw_experimental_components = components.OfType <Component>().ToList();
TopDownProteoform td1 = ConstructorsForTesting.TopDownProteoform("ACCESSION_1", 1000.0, 45);
TopDownProteoform td2 = ConstructorsForTesting.TopDownProteoform("ACCESSION_2", 1001.0, 85);
TopDownProteoform td3 = ConstructorsForTesting.TopDownProteoform("ACCESSION_3", 1131.04, 45);
TheoreticalProteoform t1 = ConstructorsForTesting.make_a_theoretical("ACCESSION", 1000.0, 1);
//need to make theoretical accession database
TestProteoformCommunityRelate.prepare_for_et(new List <double>()
{
0
});
Sweet.lollipop.target_proteoform_community.community_number = -100;
Sweet.lollipop.theoretical_database.theoreticals_by_accession = new Dictionary <int, Dictionary <string, List <TheoreticalProteoform> > >();
Sweet.lollipop.theoretical_database.theoreticals_by_accession.Add(-100, new Dictionary <string, List <TheoreticalProteoform> >());
Sweet.lollipop.theoretical_database.theoreticals_by_accession[-100].Add(t1.accession, new List <TheoreticalProteoform>()
{
t1
});
//need to make decon error top "deconvolution error"
ModificationMotif motif;
ModificationMotif.TryGetMotif("S", out motif);
Modification m = new Modification("id", _modificationType: "modtype", _target: motif, _locationRestriction: "Anywhere.", _monoisotopicMass: 1);
Sweet.lollipop.theoretical_database.all_mods_with_mass.Add(m);
PtmSet set = new PtmSet(new List <Ptm> {
new Ptm(-1, m)
});
Sweet.lollipop.theoretical_database.all_possible_ptmsets.Add(set);
Sweet.lollipop.modification_ranks.Add(-1.0023, 2);
Sweet.lollipop.theoretical_database.possible_ptmset_dictionary.Add(-1.0, new List <PtmSet>()
{
set
});
//need missing error
Modification m2 = new Modification("id", _modificationType: "modtype", _target: motif, _locationRestriction: "Anywhere.", _monoisotopicMass: 1);
Sweet.lollipop.theoretical_database.all_mods_with_mass.Add(m2);
PtmSet set2 = new PtmSet(new List <Ptm> {
new Ptm(-1, m2)
});
Sweet.lollipop.theoretical_database.all_possible_ptmsets.Add(set2);
Sweet.lollipop.modification_ranks.Add(-87.03, 2);
Sweet.lollipop.theoretical_database.possible_ptmset_dictionary.Add(-87.0, new List <PtmSet>()
{
set2
});
Sweet.lollipop.target_proteoform_community.theoretical_proteoforms = new List <TheoreticalProteoform>()
{
t1
}.ToArray();
Sweet.lollipop.topdown_proteoforms = new List <TopDownProteoform> {
td1, td2, td3
};
Sweet.lollipop.add_td_proteoforms = true;
Sweet.lollipop.aggregate_proteoforms(Sweet.lollipop.validate_proteoforms, Sweet.lollipop.raw_neucode_pairs, Sweet.lollipop.raw_experimental_components, Sweet.lollipop.raw_quantification_components, 0);
List <ProteoformRelation> relations = Sweet.lollipop.target_proteoform_community.relate(Sweet.lollipop.target_proteoform_community.experimental_proteoforms, Sweet.lollipop.target_proteoform_community.theoretical_proteoforms, ProteoformComparison.ExperimentalTheoretical, true, Environment.CurrentDirectory, true);
List <DeltaMassPeak> peaks = Sweet.lollipop.target_proteoform_community.accept_deltaMass_peaks(relations, new Dictionary <string, List <ProteoformRelation> >());
//should have 4 experimental proteoforms -- 3 topdown, 1 not topdown experimental
Assert.AreEqual(3, Sweet.lollipop.target_proteoform_community.experimental_proteoforms.Count(e => e.topdown_id));
Assert.AreEqual(4, Sweet.lollipop.target_proteoform_community.experimental_proteoforms.Count());
Assert.AreEqual(1, relations.Count);
Assert.AreEqual(1, relations.Count(r => r.RelationType == ProteoformComparison.ExperimentalTheoretical && (r.connected_proteoforms[0] as ExperimentalProteoform).topdown_id));
Assert.AreEqual(1, td1.relationships.Count(r => r.RelationType == ProteoformComparison.ExperimentalTheoretical && (r.connected_proteoforms[0] as ExperimentalProteoform).topdown_id));
Assert.AreEqual(0, td2.relationships.Count());
Assert.AreEqual(0, td3.relationships.Count());
//accession 3 has higher score... gets td
Sweet.lollipop.clear_td();
td3 = ConstructorsForTesting.TopDownProteoform("ACCESSION_3", 1000.0, 45);
TopDownHit h3 = new TopDownHit();
h3.score = 100;
td3.topdown_hits = new List <TopDownHit>()
{
h3
};
TopDownHit h4 = new TopDownHit();
h4.score = 1;
TopDownProteoform td4 = ConstructorsForTesting.TopDownProteoform("ACCESSION_4", 1001.0, 45);
td4.topdown_hits = new List <TopDownHit>()
{
h4
};
Sweet.lollipop.topdown_proteoforms = new List <TopDownProteoform> {
td4, td3
};
Sweet.lollipop.topdown_proteoforms.OrderBy(p => p.modified_mass);
Sweet.lollipop.aggregate_proteoforms(Sweet.lollipop.validate_proteoforms, Sweet.lollipop.raw_neucode_pairs, Sweet.lollipop.raw_experimental_components, Sweet.lollipop.raw_quantification_components, 0);
Assert.AreEqual(0, Math.Round(td3.modified_mass - td3.matching_experimental.modified_mass, 0));
Assert.IsNull(td4.matching_experimental);
}