private bool assign_pf_identity(ExperimentalProteoform e, PtmSet set, int begin, int end, ProteoformRelation r, TheoreticalProteoform theoretical_base, List <Proteoform> linked_proteoform_references, bool check_ambiguous_IDs) { bool identification_assigned = false; if (!Sweet.lollipop.id_use_ppm_tolerance || Math.Abs(e.calculate_mass_error(theoretical_base, set, begin, end) * 1e6 / e.modified_mass) < Sweet.lollipop.id_ppm_tolerance) { int new_begin = begin; int new_end = end; PtmSet new_set = new PtmSet(new List <Ptm>(set.ptm_combination)); List <Ptm> remove = new List <Ptm>(); //do retention of M first foreach (var mod in new_set.ptm_combination.Where(m => m.modification.ModificationType == "AminoAcid")) { new_begin--; remove.Add(mod); } foreach (var mod in new_set.ptm_combination.Where(m => m.modification.ModificationType == "Missing")) { if (!new_set.ptm_combination.Any(m => m.modification.ModificationType == "AminoAcid") && begin >= theoretical_base.begin) { if (theoretical_base.sequence[begin - theoretical_base.begin].ToString() == mod.modification.Target.ToString()) { new_begin++; remove.Add(mod); //dont have in ptmset --> change the begin & end } } if (!remove.Contains(mod) && theoretical_base.sequence[end - theoretical_base.begin].ToString() == mod.modification.Target.ToString()) { new_end--; remove.Add(mod); } } foreach (var ptm in remove) { new_set.ptm_combination.Remove(ptm); } new_set = new PtmSet(new_set.ptm_combination); if (e.linked_proteoform_references == null) { identification_assigned = true; if (linked_proteoform_references != null) { e.linked_proteoform_references = new List <Proteoform>(linked_proteoform_references); e.linked_proteoform_references.Add(this); } else { e.linked_proteoform_references = new List <Proteoform>() { theoretical_base }; } e.relation_to_id = r; e.ptm_set = new_set; e.begin = new_begin; e.end = new_end; if (e.gene_name == null) { e.gene_name = theoretical_base.gene_name; } else { e.gene_name.gene_names.Concat(this.gene_name.gene_names); } } else { if (linked_proteoform_references != null && !linked_proteoform_references.Contains(e)) { bool different_id = e.gene_name.get_prefered_name(Lollipop.preferred_gene_label) != theoretical_base.gene_name.get_prefered_name(Lollipop.preferred_gene_label) || ExperimentalProteoform.get_sequence(e.linked_proteoform_references.First() as TheoreticalProteoform, e.begin, e.end) != ExperimentalProteoform.get_sequence(theoretical_base, new_begin, new_end) || !e.ptm_set.same_ptmset(new_set, true); List <Modification> this_known_mods = theoretical_base.ExpandedProteinList.SelectMany(p => p.OneBasedPossibleLocalizedModifications).SelectMany(kv => kv.Value).Where(v => v.MonoisotopicMass != 0).ToList(); List <Modification> previous_id_known_mods = (e.linked_proteoform_references.First() as TheoreticalProteoform).ExpandedProteinList.SelectMany(p => p.OneBasedPossibleLocalizedModifications).SelectMany(kv => kv.Value).Where(v => v.MonoisotopicMass != 0).ToList(); if (!Sweet.lollipop.topdown_theoretical_reduce_ambiguity || (theoretical_base.topdown_theoretical && !(e.linked_proteoform_references.First() as TheoreticalProteoform).topdown_theoretical)) { if (!Sweet.lollipop.annotated_PTMs_reduce_ambiguity || (new_set.ptm_combination.All(mod1 => modification_is_adduct(mod1.modification) || this_known_mods.Select(mod2 => UnlocalizedModification.LookUpId(mod2)).Contains(UnlocalizedModification.LookUpId(mod1.modification))) && !e.ptm_set.ptm_combination.All(mod1 => modification_is_adduct(mod1.modification) || previous_id_known_mods.Select(mod2 => UnlocalizedModification.LookUpId(mod2)).Contains(UnlocalizedModification.LookUpId(mod1.modification))))) { if (Sweet.lollipop.topdown_theoretical_reduce_ambiguity || Sweet.lollipop.annotated_PTMs_reduce_ambiguity) { if (Sweet.lollipop.remove_bad_connections && different_id) //&& e.relation_to_id != r) { e.relation_to_id.Identification = false; e.relation_to_id.represented_ptmset = null; } e.linked_proteoform_references = null; e.ptm_set = new PtmSet(new List <Ptm>()); e.begin = 0; e.end = 0; e.gene_name = null; e.ambiguous_identifications.Clear(); ProteoformRelation relation = null; e.relation_to_id = relation; //reassign the topdown - based ID return(this.assign_pf_identity(e, set, begin, end, r, theoretical_base, linked_proteoform_references, true)); } } } if (Sweet.lollipop.topdown_theoretical_reduce_ambiguity && (e.linked_proteoform_references.First() as TheoreticalProteoform).topdown_theoretical && !theoretical_base.topdown_theoretical) { } else if (Sweet.lollipop.annotated_PTMs_reduce_ambiguity && !new_set.ptm_combination.All(mod1 => modification_is_adduct(mod1.modification) || this_known_mods.Select(mod2 => UnlocalizedModification.LookUpId(mod2)).Contains(UnlocalizedModification.LookUpId(mod1.modification))) && e.ptm_set.ptm_combination.All(mod1 => modification_is_adduct(mod1.modification) || previous_id_known_mods.Select(mod2 => UnlocalizedModification.LookUpId(mod2)).Contains(UnlocalizedModification.LookUpId(mod1.modification)))) { } else { if (different_id) { var new_linked_proteoform_references = new List <Proteoform>(linked_proteoform_references); new_linked_proteoform_references.Add(this); AmbiguousIdentification new_id = new AmbiguousIdentification(new_begin, new_end, new_set, r, theoretical_base, new_linked_proteoform_references); lock (e.ambiguous_identifications) { if (!e.ambiguous_identifications.Any(p => p.theoretical_base.gene_name.primary == new_id.theoretical_base.gene_name.primary && ExperimentalProteoform.get_sequence(p.theoretical_base, p.begin, p.end) == ExperimentalProteoform.get_sequence(new_id.theoretical_base, new_id.begin, new_id.end) && p.ptm_set.same_ptmset(new_id.ptm_set, true))) { e.ambiguous_identifications.Add(new_id); identification_assigned = true; } } } } } } } if (check_ambiguous_IDs) { //remove bad relations if using td to reduce ambiguity if (identification_assigned) { List <AmbiguousIdentification> to_remove = new List <AmbiguousIdentification>(); List <Modification> previous_id_known_mods = (e.linked_proteoform_references.First() as TheoreticalProteoform).ExpandedProteinList.SelectMany(p => p.OneBasedPossibleLocalizedModifications).SelectMany(kv => kv.Value).Where(m => m.MonoisotopicMass != 0).ToList(); if (theoretical_base.topdown_theoretical && Sweet.lollipop.topdown_theoretical_reduce_ambiguity) { to_remove.AddRange(e.ambiguous_identifications.Where(id => !id.theoretical_base.topdown_theoretical)); } if (Sweet.lollipop.annotated_PTMs_reduce_ambiguity && e.ptm_set.ptm_combination.All(mod1 => modification_is_adduct(mod1.modification) || previous_id_known_mods.Select(mod2 => UnlocalizedModification.LookUpId(mod2)).Contains(UnlocalizedModification.LookUpId(mod1.modification)))) { foreach (var ambiguous_id in e.ambiguous_identifications) { List <Modification> ambiguous_id_known_mods = ambiguous_id.theoretical_base.ExpandedProteinList.SelectMany(p => p.OneBasedPossibleLocalizedModifications).SelectMany(kv => kv.Value).Where(m => m.MonoisotopicMass != 0).ToList(); if (ambiguous_id.ptm_set.ptm_combination.Any(mod1 => !modification_is_adduct(mod1.modification) && !ambiguous_id_known_mods.Select(mod2 => UnlocalizedModification.LookUpId(mod2)).Contains(UnlocalizedModification.LookUpId(mod1.modification)))) { to_remove.Add(ambiguous_id); } } } foreach (var x in to_remove) { if (e.ambiguous_identifications.Contains(x)) { e.ambiguous_identifications.Remove(x); if (Sweet.lollipop.remove_bad_connections) { if (e.relation_to_id != x.relation) { x.relation.Identification = false; x.relation.represented_ptmset = null; } } } } foreach (var x in e.ambiguous_identifications) { x.relation.Identification = true; } } if (this as ExperimentalProteoform != null && (this as ExperimentalProteoform).ambiguous_identifications.Count > 0) { lock ((this as ExperimentalProteoform).ambiguous_identifications) { int count = (this as ExperimentalProteoform).ambiguous_identifications.Count; PtmSet[] new_ptm_set = new PtmSet[count]; Parallel.For(0, count, i => { var id = (this as ExperimentalProteoform).ambiguous_identifications[i]; new_ptm_set[i] = determine_mod_change(e, this, id.theoretical_base, r, id.ptm_set, id.begin, id.end); }); for (int i = 0; i < count; i++) { if (new_ptm_set[i] != null) { var id = (this as ExperimentalProteoform).ambiguous_identifications[i]; if (assign_pf_identity(e, new_ptm_set[i], id.begin, id.end, r, id.theoretical_base, id.linked_proteoform_references, false)) { identification_assigned = true; } } } } } } return(identification_assigned); }
public void identify_experimentals() { HashSet <ExperimentalProteoform> identified_experimentals = new HashSet <ExperimentalProteoform>(); if (Sweet.lollipop.identify_from_td_nodes) { foreach (TopDownProteoform topdown in experimental_proteoforms.Where(e => e.topdown_id)) { Sweet.lollipop.theoretical_database .theoreticals_by_accession[Sweet.lollipop.target_proteoform_community.community_number] .TryGetValue(topdown.accession.Split('_')[0].Split('-')[0], out var t); if (t != null && t.Count > 0) { TheoreticalProteoform theoretical = new TheoreticalProteoform(topdown.accession, topdown.name, topdown.sequence, t.First().ExpandedProteinList, topdown.modified_mass, topdown.lysine_count, topdown.topdown_ptm_set, true, false, null); theoretical.topdown_theoretical = true; theoretical.new_topdown_proteoform = true; theoretical.begin = topdown.topdown_begin; theoretical.end = topdown.topdown_end; foreach (ExperimentalProteoform e in topdown.identify_connected_experimentals(theoretical, topdown.topdown_begin, topdown.topdown_end, new PtmSet(topdown.topdown_ptm_set.ptm_combination), null)) { identified_experimentals.Add(e); } } } } foreach (TheoreticalProteoform t in theoretical_proteoforms.OrderBy(t => t.topdown_theoretical)) { lock (identified_experimentals) foreach (ExperimentalProteoform e in t.identify_connected_experimentals(t, t.begin, t.end, t.ptm_set, t.linked_proteoform_references)) { identified_experimentals.Add(e); } } //Continue looking for new experimental identifications until no more remain to be identified List <ExperimentalProteoform> newly_identified_experimentals = new List <ExperimentalProteoform>(identified_experimentals).OrderBy(p => p.relationships.Count(r => r.candidate_ptmset != null) > 0 ? p.relationships.Where(r => r.candidate_ptmset != null).Min(r => Math.Abs(r.DeltaMass - r.candidate_ptmset.mass)) : 1e6).ThenBy(p => p.modified_mass).ToList(); int last_identified_count = identified_experimentals.Count - 1; while (newly_identified_experimentals.Count > 0) //&& identified_experimentals.Count > last_identified_count) { last_identified_count = identified_experimentals.Count; HashSet <ExperimentalProteoform> tmp_new_experimentals = new HashSet <ExperimentalProteoform>(); foreach (ExperimentalProteoform id_experimental in newly_identified_experimentals) { { lock (identified_experimentals) lock (tmp_new_experimentals) foreach (ExperimentalProteoform new_e in id_experimental.identify_connected_experimentals(id_experimental.linked_proteoform_references.First() as TheoreticalProteoform, id_experimental.begin, id_experimental.end, id_experimental.ptm_set, id_experimental.linked_proteoform_references)) { identified_experimentals.Add(new_e); tmp_new_experimentals.Add(new_e); } } } newly_identified_experimentals = new List <ExperimentalProteoform>(tmp_new_experimentals); } List <string> topdown_ids = Sweet.lollipop.topdown_proteoforms .Select(p => p.accession.Split('_')[0].Split('-')[0] + "_" + p.sequence + "_" + string.Join(", ", p.topdown_ptm_set.ptm_combination.Where(m => m.modification.ModificationType != "Deconvolution Error").Select(ptm => UnlocalizedModification.LookUpId(ptm.modification)).OrderBy(m => m))).ToList(); //determine identified experimentals that are adducts //checks if any experimentals have same mods as e's ptmset, except e has additional adduct only mods. Parallel.ForEach(experimental_proteoforms, e => { e.adduct = e.linked_proteoform_references != null && e.ptm_set.ptm_combination.Any(m => Proteoform.modification_is_adduct(m.modification)) && experimental_proteoforms.Any(l => l.linked_proteoform_references != null && l.gene_name.get_prefered_name(Lollipop.preferred_gene_label) == e.gene_name.get_prefered_name(Lollipop.preferred_gene_label) && l.ptm_set.ptm_combination.Count < e.ptm_set.ptm_combination.Count && e.ptm_set.ptm_combination.Where(m => l.ptm_set.ptm_combination.Count(p => UnlocalizedModification.LookUpId(p.modification) == UnlocalizedModification.LookUpId(m.modification)) != e.ptm_set.ptm_combination.Count(p => UnlocalizedModification.LookUpId(p.modification) == UnlocalizedModification.LookUpId(m.modification))) .Count(p => !Proteoform.modification_is_adduct(p.modification)) == 0 ); if (e as TopDownProteoform != null) { (e as TopDownProteoform).set_correct_id(); } if (e.linked_proteoform_references != null) { var mods = e.ptm_set.ptm_combination.Where(p => !Proteoform.modification_is_adduct(p.modification)) .Select(ptm => UnlocalizedModification.LookUpId(ptm.modification)).ToList().Distinct().OrderBy(m => m).ToList(); e.uniprot_mods = ""; string add = ""; foreach (string mod in mods) { // positions with mod List <int> theo_ptms = (e.linked_proteoform_references.First() as TheoreticalProteoform).ExpandedProteinList.SelectMany(p => p .OneBasedPossibleLocalizedModifications) .Where(p => p.Key >= e.begin && p.Key <= e.end && p.Value.Select(m => UnlocalizedModification.LookUpId(m)).Contains(mod)) .Select(m => m.Key).ToList(); if (theo_ptms.Count > 0) { add += mod + " @ " + string.Join(", ", theo_ptms) + "; "; } if (e.ptm_set.ptm_combination.Where(ptm => ptm.modification.ModificationType != "Deconvolution Error" && !Proteoform.modification_is_adduct(ptm.modification)).Select(ptm => UnlocalizedModification.LookUpId(ptm.modification)) .Count(m => m == mod) > theo_ptms.Count) { e.novel_mods = true; } } e.uniprot_mods += add; if (add.Length == 0) { e.uniprot_mods += "N/A"; } foreach (var ambig_id in e.ambiguous_identifications) { var ambig_mods = ambig_id.ptm_set.ptm_combination.Where(p => !Proteoform.modification_is_adduct(p.modification)) .Select(ptm => UnlocalizedModification.LookUpId(ptm.modification)).ToList().Distinct().OrderBy(m => m).ToList(); e.uniprot_mods += " | "; add = ""; foreach (var mod in ambig_mods) { // positions with mod List <int> theo_ptms = ambig_id.theoretical_base.ExpandedProteinList.SelectMany(p => p .OneBasedPossibleLocalizedModifications) .Where(p => p.Key >= ambig_id.begin && p.Key <= ambig_id.end && p.Value.Select(m => UnlocalizedModification.LookUpId(m)).Contains(mod)) .Select(m => m.Key).ToList(); if (theo_ptms.Count > 0) { add += mod + " @ " + string.Join(", ", theo_ptms) + "; "; } if (ambig_id.ptm_set.ptm_combination.Where(ptm => ptm.modification.ModificationType != "Deconvolution Error" && !Proteoform.modification_is_adduct(ptm.modification)).Select(ptm => UnlocalizedModification.LookUpId(ptm.modification)) .Count(m => m == mod) > theo_ptms.Count) { e.novel_mods = true; } } e.uniprot_mods += add; if (add.Length == 0) { e.uniprot_mods += "N/A"; } } } //determine level # e.proteoform_level_description = ""; if (e.linked_proteoform_references == null) { e.proteoform_level = 5; e.proteoform_level_description = "Unidentified"; } else if (e.ambiguous_identifications.Count == 0) { if (e.ptm_set.ptm_combination.Count == 0) { e.proteoform_level = 1; } else { e.proteoform_level = 2; e.proteoform_level_description += "PTM localization ambiguity; "; } //check if accessions had been grouped in constructing the theoretical database if ((e.linked_proteoform_references.First() as TheoreticalProteoform).ExpandedProteinList.SelectMany(p => p.AccessionList).Select(a => a.Split('_')[0]).Distinct().Count() > 1) { e.proteoform_level += 1; e.proteoform_level_description += "Gene ambiguity; "; } } else { var unique_accessions = new List <string>() { e.linked_proteoform_references.First().accession.Split('_')[0].Split('-')[0] }.Concat(e.ambiguous_identifications.Select(a => a.theoretical_base.accession.Split('_')[0].Split('-')[0])).Distinct(); var unique_sequences = new List <string>() { ExperimentalProteoform.get_sequence(e.linked_proteoform_references.First() as TheoreticalProteoform, e.begin, e.end) }. Concat(e.ambiguous_identifications.Select(a => ExperimentalProteoform.get_sequence(a.theoretical_base, a.begin, a.end))).Distinct(); var unique_PTMs = new List <string>() { string.Join(", ", e.ptm_set.ptm_combination.Where(m => m.modification.ModificationType != "Deconvolution Error").Select(ptm => UnlocalizedModification.LookUpId(ptm.modification)).OrderBy(m => m)) }.Concat(e.ambiguous_identifications.Select(a => string.Join(", ", a.ptm_set.ptm_combination.Where(m => m.modification.ModificationType != "Deconvolution Error").Select(ptm => UnlocalizedModification.LookUpId(ptm.modification))))).Distinct(); int gene_ambiguity = unique_accessions.Count() > 1 ? 1 : 0; //check if accessions had been grouped in constructing the theoretical database if ((e.linked_proteoform_references.First() as TheoreticalProteoform).ExpandedProteinList.SelectMany(p => p.AccessionList).Select(a => a.Split('_')[0]).Distinct().Count() > 1) { gene_ambiguity = 1; } int sequence_ambiguity = unique_sequences.Count() > 1 ? 1 : 0; int PTM_ambiguity = unique_PTMs.Count() > 1 ? 1 : 0; int PTM_location = e.ptm_set.ptm_combination.Count(m => m.modification.ModificationType != "Deconvolution Error") > 0 || e.ambiguous_identifications.Any(a => a.ptm_set.ptm_combination.Count(m => m.modification.ModificationType != "Deconvolution Error") > 0) ? 1 : 0; e.proteoform_level = 1 + gene_ambiguity + sequence_ambiguity + PTM_ambiguity + PTM_location; if (gene_ambiguity > 0) { e.proteoform_level_description += "Gene ambiguity; "; } if (sequence_ambiguity > 0) { e.proteoform_level_description += "Sequence ambiguity; "; } if (PTM_ambiguity > 0) { e.proteoform_level_description += "PTM identity ambiguity; "; } if (PTM_location > 0) { e.proteoform_level_description += "PTM localization ambiguity; "; } } if (e.proteoform_level == 1) { e.proteoform_level_description = "Unambiguous"; } //determine if new intact-mass ID e.new_intact_mass_id = false; if (!e.topdown_id && e.linked_proteoform_references != null && e.ambiguous_identifications.Count == 0) { string this_id = string.Join(",", (e.linked_proteoform_references.First() as TheoreticalProteoform).ExpandedProteinList.SelectMany(p => p.AccessionList.Select(a => a.Split('_')[0])).Distinct()) + "_" + ExperimentalProteoform.get_sequence(e.linked_proteoform_references.First() as TheoreticalProteoform, e.begin, e.end) + "_" + string.Join(", ", e.ptm_set.ptm_combination.Where(m => m.modification.ModificationType != "Deconvolution Error").Select(ptm => UnlocalizedModification.LookUpId(ptm.modification)).OrderBy(m => m)); if (!topdown_ids.Any(t => this_id.Split('_')[0].Split(',').Contains(t.Split('_')[0]) && this_id.Split('_')[1] == t.Split('_')[1] && this_id.Split('_')[2] == t.Split('_')[2])) { e.new_intact_mass_id = true; } } }); if (Sweet.lollipop.remove_bad_connections) { if (theoretical_proteoforms.Count > 0 || (Sweet.lollipop.identify_from_td_nodes && experimental_proteoforms.Count(e => e.topdown_id) > 0)) { Parallel.ForEach(relations, r => { r.Accepted = r.Identification; }); } } }