static void Main(string[] args)
{
// the indexes of data, contacts1 and contacts2 all match
var data = MultiBindingInterface.Load(@"c:\pdb\new_data_set.csv");
var contactsPartner1 =
data.Select(
a =>
{
var x = FindAtomicContacts.AtomPair.LoadAtomPairList(@"C:\pdb\new_data_set\contacts\contacts_" +
a.InteractionChainsPdb1.ToLowerInvariant() + ".pdb")
.Where(
b =>
(b.Atom1.chainID.FieldValue.ToUpperInvariant() ==
a.InteractionChainsPdb1Chain1.ToUpperInvariant()
&&
b.Atom2.chainID.FieldValue.ToUpperInvariant() ==
a.InteractionChainsPdb1Chain2.ToUpperInvariant())
||
(b.Atom1.chainID.FieldValue.ToUpperInvariant() ==
a.InteractionChainsPdb1Chain2.ToUpperInvariant()
&&
b.Atom2.chainID.FieldValue.ToUpperInvariant() ==
a.InteractionChainsPdb1Chain1.ToUpperInvariant())).ToList();
x = x.Select(c =>
{
if (c.Atom1.chainID.FieldValue.ToUpperInvariant() == a.InteractionChainsPdb1Chain2.ToUpperInvariant())
{
c.SwapAtoms();
}
return(c);
}).ToList();
return(x);
}).ToList();
var contactsPartner2 =
data.Select(
a =>
{
var x = FindAtomicContacts.AtomPair.LoadAtomPairList(@"C:\pdb\new_data_set\contacts\contacts_" +
a.InteractionChainsPdb2.ToLowerInvariant() + ".pdb")
.Where(
b =>
(b.Atom1.chainID.FieldValue.ToUpperInvariant() ==
a.InteractionChainsPdb2Chain1.ToUpperInvariant()
&&
b.Atom2.chainID.FieldValue.ToUpperInvariant() ==
a.InteractionChainsPdb2Chain2.ToUpperInvariant())
||
(b.Atom1.chainID.FieldValue.ToUpperInvariant() ==
a.InteractionChainsPdb2Chain2.ToUpperInvariant()
&&
b.Atom2.chainID.FieldValue.ToUpperInvariant() ==
a.InteractionChainsPdb2Chain1.ToUpperInvariant())).ToList();
x = x.Select(c =>
{
if (c.Atom1.chainID.FieldValue.ToUpperInvariant() == a.InteractionChainsPdb2Chain2.ToUpperInvariant())
{
c.SwapAtoms();
}
return(c);
}).ToList();
return(x);
}).ToList();
var interfacePartner1 = contactsPartner1.Select(a =>
{
var resSeqChain1 = a.Select(b => int.Parse(b.Atom1.resSeq.FieldValue)).ToList();
var resSeqChain2 = a.Select(b => int.Parse(b.Atom2.resSeq.FieldValue)).ToList();
if (resSeqChain1.Count > 0 && resSeqChain2.Count > 0)
{
return(new Tuple <int, int, int, int>(resSeqChain1.Min(), resSeqChain1.Max(), resSeqChain2.Min(),
resSeqChain2.Max()));
}
else
{
return(null);
}
}).ToList();
var interfacePartner2 = contactsPartner2.Select(a =>
{
var resSeqChain1 = a.Select(b => int.Parse(b.Atom1.resSeq.FieldValue)).ToList();
var resSeqChain2 = a.Select(b => int.Parse(b.Atom2.resSeq.FieldValue)).ToList();
if (resSeqChain1.Count > 0 && resSeqChain2.Count > 0)
{
return(new Tuple <int, int, int, int>(resSeqChain1.Min(), resSeqChain1.Max(), resSeqChain2.Min(),
resSeqChain2.Max()));
}
else
{
return(null);
}
}).ToList();
var resultData = new List <string>();
for (int index = 0; index < data.Count; index++)
{
var d = data[index];
var cp1 = contactsPartner1[index];
var cp2 = contactsPartner2[index];
var ip1 = interfacePartner1[index];
var ip2 = interfacePartner2[index];
if (d == null || cp1 == null || cp2 == null || ip1 == null || ip2 == null)
{
continue;
}
if (cp1.Count == 0 || cp2.Count == 0)
{
continue;
}
var p1c1_pdb = FindAtomicContacts.PdbAtomicChains(@"c:\pdb\new_data_set\" + d.InteractionChainsPdb1 + ".pdb", new string[] { d.InteractionChainsPdb1Chain1 }, -1, -1, true);
var p1c2_pdb = FindAtomicContacts.PdbAtomicChains(@"c:\pdb\new_data_set\" + d.InteractionChainsPdb1 + ".pdb", new string[] { d.InteractionChainsPdb1Chain2 }, -1, -1, true);
var p2c1_pdb = FindAtomicContacts.PdbAtomicChains(@"c:\pdb\new_data_set\" + d.InteractionChainsPdb2 + ".pdb", new string[] { d.InteractionChainsPdb2Chain1 }, -1, -1, true);
var p2c2_pdb = FindAtomicContacts.PdbAtomicChains(@"c:\pdb\new_data_set\" + d.InteractionChainsPdb2 + ".pdb", new string[] { d.InteractionChainsPdb2Chain2 }, -1, -1, true);
var p1c1_res_seq = p1c1_pdb.ChainList.First().AtomList.Select(a => int.Parse(a.resSeq.FieldValue)).ToList();
var p1c2_res_seq = p1c2_pdb.ChainList.First().AtomList.Select(a => int.Parse(a.resSeq.FieldValue)).ToList();
var p2c1_res_seq = p2c1_pdb.ChainList.First().AtomList.Select(a => int.Parse(a.resSeq.FieldValue)).ToList();
var p2c2_res_seq = p2c2_pdb.ChainList.First().AtomList.Select(a => int.Parse(a.resSeq.FieldValue)).ToList();
var cp1a1_res_seq = cp1.Select(a => int.Parse(a.Atom1.resSeq.FieldValue)).ToList();
var cp1a2_res_seq = cp1.Select(a => int.Parse(a.Atom2.resSeq.FieldValue)).ToList();
var cp2a1_res_seq = cp2.Select(a => int.Parse(a.Atom1.resSeq.FieldValue)).ToList();
var cp2a2_res_seq = cp2.Select(a => int.Parse(a.Atom2.resSeq.FieldValue)).ToList();
var cp1a1_min = cp1a1_res_seq.Min();
var cp1a1_max = cp1a1_res_seq.Max();
var cp1a2_min = cp1a2_res_seq.Min();
var cp1a2_max = cp1a2_res_seq.Max();
var cp2a1_min = cp2a1_res_seq.Min();
var cp2a1_max = cp2a1_res_seq.Max();
var cp2a2_min = cp2a2_res_seq.Min();
var cp2a2_max = cp2a2_res_seq.Max();
var cp1a1_best50_min = int.MinValue;
var cp1a1_best50_max = int.MinValue;
var cp1a1_best50_interactions = int.MinValue;
var cp1a1_best50_middle_finder = new List <Tuple <int, int, int> >();
var interface_target_length = 50;
for (var x = cp1a1_min - interface_target_length; x <= cp1a1_max; x++)
{
if (Math.Abs(cp1a1_max - cp1a1_min) <= interface_target_length)
{
cp1a1_best50_min = cp1a1_min;
cp1a1_best50_max = cp1a1_max;
cp1a1_best50_interactions = cp1a1_res_seq.Count;
break;
}
var min = x;
var max = x + interface_target_length > cp1a1_max ? cp1a1_max : x + interface_target_length;
var best50 = cp1a1_res_seq.Count(a => a >= cp1a1_best50_min && a <= cp1a1_best50_max);
if (best50 == cp1a1_best50_interactions)
{
cp1a1_best50_middle_finder.Add(new Tuple <int, int, int>(min, max, best50));
}
if (cp1a1_best50_interactions == int.MinValue || best50 > cp1a1_best50_interactions)
{
cp1a1_best50_middle_finder.Clear();
cp1a1_best50_middle_finder.Add(new Tuple <int, int, int>(min, max, best50));
cp1a1_best50_min = min;
cp1a1_best50_max = max;
cp1a1_best50_interactions = best50;
}
if (x + interface_target_length >= cp1a1_max)
{
break;
}
}
if (cp1a1_best50_middle_finder.Count > 2)
{
var middle = cp1a1_best50_middle_finder[cp1a1_best50_middle_finder.Count / 2];
cp1a1_best50_min = middle.Item1;
cp1a1_best50_max = middle.Item2;
cp1a1_best50_interactions = middle.Item3;
}
var cp1a2_best50_min = int.MinValue;
var cp1a2_best50_max = int.MinValue;
var cp1a2_best50_interactions = int.MinValue;
var cp1a2_best50_middle_finder = new List <Tuple <int, int, int> >();
for (var x = cp1a2_min - interface_target_length; x <= cp1a2_max; x++)
{
if (Math.Abs(cp1a2_max - cp1a2_min) <= interface_target_length)
{
cp1a2_best50_min = cp1a2_min;
cp1a2_best50_max = cp1a2_max;
cp1a2_best50_interactions = cp1a2_res_seq.Count;
break;
}
var min = x;
var max = x + interface_target_length > cp1a2_max ? cp1a2_max : x + interface_target_length;
var best50 = cp1a2_res_seq.Count(a => a >= cp1a2_best50_min && a <= cp1a2_best50_max);
if (best50 == cp1a2_best50_interactions)
{
cp1a2_best50_middle_finder.Add(new Tuple <int, int, int>(min, max, best50));
}
if (cp1a2_best50_interactions == int.MinValue || best50 > cp1a2_best50_interactions)
{
cp1a2_best50_middle_finder.Clear();
cp1a2_best50_middle_finder.Add(new Tuple <int, int, int>(min, max, best50));
cp1a2_best50_min = min;
cp1a2_best50_max = max;
cp1a2_best50_interactions = best50;
}
if (x + interface_target_length >= cp1a2_max)
{
break;
}
}
if (cp1a2_best50_middle_finder.Count > 2)
{
var middle = cp1a2_best50_middle_finder[cp1a2_best50_middle_finder.Count / 2];
cp1a2_best50_min = middle.Item1;
cp1a2_best50_max = middle.Item2;
cp1a2_best50_interactions = middle.Item3;
}
var cp2a1_best50_min = int.MinValue;
var cp2a1_best50_max = int.MinValue;
var cp2a1_best50_interactions = int.MinValue;
var cp2a1_best50_middle_finder = new List <Tuple <int, int, int> >();
for (var x = cp2a1_min - interface_target_length; x <= cp2a1_max; x++)
{
if (Math.Abs(cp2a1_max - cp2a1_min) <= interface_target_length)
{
cp2a1_best50_min = cp2a1_min;
cp2a1_best50_max = cp2a1_max;
cp2a1_best50_interactions = cp2a1_res_seq.Count;
break;
}
var min = x;
var max = x + interface_target_length > cp2a1_max ? cp2a1_max : x + interface_target_length;
var best50 = cp2a1_res_seq.Count(a => a >= cp2a1_best50_min && a <= cp2a1_best50_max);
if (best50 == cp2a1_best50_interactions)
{
cp2a1_best50_middle_finder.Add(new Tuple <int, int, int>(min, max, best50));
}
if (cp2a1_best50_interactions == int.MinValue || best50 > cp2a1_best50_interactions)
{
cp2a1_best50_middle_finder.Clear();
cp2a1_best50_middle_finder.Add(new Tuple <int, int, int>(min, max, best50));
cp2a1_best50_min = min;
cp2a1_best50_max = max;
cp2a1_best50_interactions = best50;
}
if (x + interface_target_length >= cp2a1_max)
{
break;
}
}
if (cp2a1_best50_middle_finder.Count > 2)
{
var middle = cp2a1_best50_middle_finder[cp2a1_best50_middle_finder.Count / 2];
cp2a1_best50_min = middle.Item1;
cp2a1_best50_max = middle.Item2;
cp2a1_best50_interactions = middle.Item3;
}
var cp2a2_best50_min = int.MinValue;
var cp2a2_best50_max = int.MinValue;
var cp2a2_best50_interactions = int.MinValue;
var cp2a2_best50_middle_finder = new List <Tuple <int, int, int> >();
for (var x = cp2a2_min - interface_target_length; x <= cp2a2_max; x++)
{
if (Math.Abs(cp2a2_max - cp2a2_min) <= interface_target_length)
{
cp2a2_best50_min = cp2a2_min;
cp2a2_best50_max = cp2a2_max;
cp2a2_best50_interactions = cp2a2_res_seq.Count;
break;
}
var min = x;
var max = x + interface_target_length > cp2a2_max ? cp2a2_max : x + interface_target_length;
var best50 = cp2a2_res_seq.Count(a => a >= cp2a2_best50_min && a <= cp2a2_best50_max);
if (best50 == cp2a2_best50_interactions)
{
cp2a2_best50_middle_finder.Add(new Tuple <int, int, int>(min, max, best50));
}
if (cp2a2_best50_interactions == int.MinValue || best50 > cp2a2_best50_interactions)
{
cp2a2_best50_middle_finder.Clear();
cp2a2_best50_middle_finder.Add(new Tuple <int, int, int>(min, max, best50));
cp2a2_best50_min = min;
cp2a2_best50_max = max;
cp2a2_best50_interactions = best50;
}
if (x + interface_target_length >= cp2a2_max)
{
break;
}
}
if (cp2a2_best50_middle_finder.Count > 2)
{
var middle = cp2a2_best50_middle_finder[cp2a2_best50_middle_finder.Count / 2];
cp2a2_best50_min = middle.Item1;
cp2a2_best50_max = middle.Item2;
cp2a2_best50_interactions = middle.Item3;
}
var cp1a1_interface = string.Join("", p1c1_pdb.ChainList.First().AtomList.Where(a =>
{
var l = int.Parse(a.resSeq.FieldValue);
return(l >= cp1a1_best50_min && l <= cp1a1_best50_max);
}).OrderBy(c => int.Parse(c.resSeq.FieldValue)).Select(b => AminoAcidConversions.AminoAcidNameToCode1L(b.resName.FieldValue)).ToList());
var cp1a2_interface = string.Join("", p1c2_pdb.ChainList.First().AtomList.Where(a =>
{
var l = int.Parse(a.resSeq.FieldValue);
return(l >= cp1a2_best50_min && l <= cp1a2_best50_max);
}).OrderBy(c => int.Parse(c.resSeq.FieldValue)).Select(b => AminoAcidConversions.AminoAcidNameToCode1L(b.resName.FieldValue)).ToList());
var cp2a1_interface = string.Join("", p2c1_pdb.ChainList.First().AtomList.Where(a =>
{
var l = int.Parse(a.resSeq.FieldValue);
return(l >= cp2a1_best50_min && l <= cp2a1_best50_max);
}).OrderBy(c => int.Parse(c.resSeq.FieldValue)).Select(b => AminoAcidConversions.AminoAcidNameToCode1L(b.resName.FieldValue)).ToList());
var cp2a2_interface = string.Join("", p2c2_pdb.ChainList.First().AtomList.Where(a =>
{
var l = int.Parse(a.resSeq.FieldValue);
return(l >= cp2a2_best50_min && l <= cp2a2_best50_max);
}).OrderBy(c => int.Parse(c.resSeq.FieldValue)).Select(b => AminoAcidConversions.AminoAcidNameToCode1L(b.resName.FieldValue)).ToList());
var cp1a1_interface_interactions = new string('_', cp1a1_interface.Length);
cp1a1_interface_interactions = string.Join("", cp1a1_interface_interactions.Select((a, i) => cp1a1_res_seq.Contains(i + cp1a1_best50_min) ? "X" : "_").ToList());
var cp1a2_interface_interactions = new string('_', cp1a2_interface.Length);
cp1a2_interface_interactions = string.Join("", cp1a2_interface_interactions.Select((a, i) => cp1a2_res_seq.Contains(i + cp1a2_best50_min) ? "X" : "_").ToList());
var cp2a1_interface_interactions = new string('_', cp2a1_interface.Length);
cp2a1_interface_interactions = string.Join("", cp2a1_interface_interactions.Select((a, i) => cp2a1_res_seq.Contains(i + cp2a1_best50_min) ? "X" : "_").ToList());
var cp2a2_interface_interactions = new string('_', cp2a2_interface.Length);
cp2a2_interface_interactions = string.Join("", cp2a2_interface_interactions.Select((a, i) => cp2a2_res_seq.Contains(i + cp2a2_best50_min) ? "X" : "_").ToList());
resultData.Add(string.Join(",", new string[] {
d.DomainSuperFamily,
d.InteractionChainsPdb1,
d.InteractionChainsPdb1Chain1, "" + ip1.Item1, "" + ip1.Item2, "" + cp1a1_best50_min, "" + cp1a1_best50_max, "" + cp1a1_best50_interactions, cp1a1_interface, cp1a1_interface_interactions,
d.InteractionChainsPdb1Chain2, "" + ip1.Item3, "" + ip1.Item4, "" + cp1a2_best50_min, "" + cp1a2_best50_max, "" + cp1a2_best50_interactions, cp1a2_interface, cp1a2_interface_interactions,
d.InteractionChainsPdb2,
d.InteractionChainsPdb2Chain1, "" + ip2.Item1, "" + ip2.Item2, "" + cp2a1_best50_min, "" + cp2a1_best50_max, "" + cp2a1_best50_interactions, cp2a1_interface, cp2a1_interface_interactions,
d.InteractionChainsPdb2Chain2, "" + ip2.Item3, "" + ip2.Item4, "" + cp2a2_best50_min, "" + cp2a2_best50_max, "" + cp2a2_best50_interactions, cp2a2_interface, cp2a2_interface_interactions,
}));
}
resultData.Insert(0, string.Join(",", new string[] {
"super family",
"partner 1 pdb id",
"partner 1 chain id 1", "p1c1 interface start", "p1c1 interface end", "p1c1 best 50 start", "p1c1 best 50 end", "p1c1 best 50 interactions", "p1c1 interface seq", "p1c1 interface mask",
"partner 1 chain id 2", "p1c2 interface start", "p1c1 interface end", "p1c2 best 50 start", "p1c2 best 50 end", "p1c2 best 50 interactions", "p1c2 interface seq", "p1c2 interface mask",
"partner 2 pdb id",
"partner 2 chain id 1", "p2c1 interface start", "p2c1 interface end", "p2c1 best 50 start", "p2c1 best 50 end", "p2c1 best 50 interactions", "p2c1 interface seq", "p2c1 interface mask",
"partner 2 chain id 2", "p2c2 interface start", "p2c2 interface end", "p2c2 best 50 start", "p2c2 best 50 end", "p2c2 best 50 interactions", "p2c2 interface seq", "p2c2 interface mask",
}));
File.WriteAllLines(@"c:\pdb\new_data_set_results.csv", resultData);
return;
}
static void Main(string[] args)
{
var pdbFolder = @"C:\ds96ub_homologs\";
var homologClusterData =
FindHomologsCluster.FindHomologsCluster.HomologClusterData.Load(
@"c:\ds96ub_homologs\ds96ub_homologs_0.7.csv");
var pdbFiles = Directory.GetFiles(pdbFolder, "*.pdb", SearchOption.TopDirectoryOnly);
var pdbIdList = pdbFiles.Select(FindAtomicContacts.PdbIdFromPdbFilename).ToList();
// only ca-atoms, ters and endmdls
var pdbAtomsText =
pdbFiles.Select(
a =>
File.ReadAllLines(a)
.Where(b => (b.StartsWith("ATOM ") && b[13] == 'C' && b[14] == 'A') || /*b.StartsWith("TER ") ||*/ b.StartsWith("ENDMDL "))
.ToList()).ToList();
// only first nmr model
pdbAtomsText = pdbAtomsText.Select(a =>
{
var x = a.FindIndex(b => b.StartsWith("ENDMDL "));
return(x == -1 ? a : a.GetRange(0, x - 1));
}).ToList();
var pdbAtoms = pdbAtomsText.Select(a => a.Select(b => new ATOM_Record(b)).ToList()).ToList();
// get list of unique chain ids
var pdbChainIds = pdbAtoms.Select((a, i) => a.Select(b => char.ToUpperInvariant(b.chainID.FieldValue[0])).ToList()).Distinct().ToList();
var pdbIdChainIdList = new List <Tuple <string, char> >();
for (var i = 0; i < pdbIdList.Count; i++)
{
pdbIdChainIdList.AddRange(pdbChainIds[i].Select(chainId => new Tuple <string, char>(pdbIdList[i], chainId)));
}
pdbIdChainIdList = pdbIdChainIdList.Distinct().ToList();
// for each chain
var pdbContacts =
pdbIdChainIdList.Select(a =>
{
var x =
FindAtomicContacts.AtomPair.LoadAtomPairList(@"C:\ds96ub_homologs\contacts\contacts_pdb" + a.Item1.ToUpperInvariant() + ".pdb")
.Where(b => char.ToUpperInvariant(b.Atom1.chainID.FieldValue[0]) == a.Item2 || char.ToUpperInvariant(b.Atom2.chainID.FieldValue[0]) == a.Item2)
.Select(c =>
{
if (char.ToUpperInvariant(c.Atom1.chainID.FieldValue[0]) != a.Item2)
{
c.SwapAtoms();
}
return(c);
}).ToList();
return(x);
}).ToList();
// res min, res max, best min, best max, interface aa, interface mask
var pdbInterfaces = new List <Ds96UbInterface>();
var interface_target_length = 50;
for (int index = 0; index < pdbContacts.Count; index++)
{
var pdbId = pdbIdChainIdList[index].Item1;
var chainId = pdbIdChainIdList[index].Item2;
var pdbContact = pdbContacts[index];
if (pdbContact.Count == 0)
{
continue;
}
var contactChains = pdbContact.Where(a => char.ToUpperInvariant(a.Atom2.chainID.FieldValue[0]) != chainId).Select(a => char.ToUpperInvariant(a.Atom2.chainID.FieldValue[0])).Distinct().ToList();
foreach (var contactChain in contactChains)
{
var pdbContactsResSeqIds =
pdbContact.Where(a => char.ToUpperInvariant(a.Atom1.chainID.FieldValue[0]) == chainId &&
char.ToUpperInvariant(a.Atom2.chainID.FieldValue[0]) == contactChain)
.Select(a => int.Parse(a.Atom1.resSeq.FieldValue))
.ToList();
var res_seq = pdbContactsResSeqIds;
var min_res_seq = pdbContactsResSeqIds.Min();
var max_res_seq = pdbContactsResSeqIds.Max();
var best50_min = int.MinValue;
var best50_max = int.MinValue;
var best50_interactions = int.MinValue;
var best50_middle_finder = new List <Tuple <int, int, int> >();
for (var x = min_res_seq - interface_target_length; x <= max_res_seq; x++)
{
if (Math.Abs(max_res_seq - min_res_seq) <= interface_target_length)
{
best50_min = min_res_seq;
best50_max = max_res_seq;
best50_interactions = res_seq.Count;
break;
}
var min = x;
var max = x + interface_target_length > max_res_seq ? max_res_seq : x + interface_target_length;
var best50 = res_seq.Count(a => a >= best50_min && a <= best50_max);
if (best50 == best50_interactions)
{
best50_middle_finder.Add(new Tuple <int, int, int>(min, max, best50));
}
if (best50_interactions == int.MinValue || best50 > best50_interactions)
{
best50_middle_finder.Clear();
best50_middle_finder.Add(new Tuple <int, int, int>(min, max, best50));
best50_min = min;
best50_max = max;
best50_interactions = best50;
}
if (x + interface_target_length >= max)
{
break;
}
}
if (best50_middle_finder.Count > 2)
{
var middle = best50_middle_finder[best50_middle_finder.Count / 2];
best50_min = middle.Item1;
best50_max = middle.Item2;
best50_interactions = middle.Item3;
}
var best50_interface_atoms = pdbAtoms[pdbIdList.IndexOf(pdbId)].Where(a =>
{
var l = int.Parse(a.resSeq.FieldValue);
var c = char.ToUpperInvariant(a.chainID.FieldValue[0]);
return(c == chainId && l >= best50_min && l <= best50_max);
}).ToList();
best50_interface_atoms = best50_interface_atoms.OrderBy(c => int.Parse(c.resSeq.FieldValue)).ToList();
var best50_interface = string.Join("", best50_interface_atoms.Select(b => AminoAcidConversions.AminoAcidNameToCode1L(b.resName.FieldValue)).ToList());
var best50_mask = new string('_', best50_interface.Length);
best50_mask = string.Join("",
best50_mask.Select((a, i) => res_seq.Contains(i + best50_min) ? "X" : "_").ToList());
pdbInterfaces.Add(new Ds96UbInterface(pdbId, chainId, contactChain, min_res_seq, max_res_seq, best50_min,
best50_max, best50_interactions, best50_interface, best50_mask, -1, "", "", 0, -1, "", "", 0));
}
}
var homologClusterIndexes = homologClusterData.Select(a => a.ClusterIndex).Distinct().ToList();
var homologClusters = homologClusterIndexes.Select(a => homologClusterData.Where(b => b.ClusterIndex == a).ToList()).ToList();
var pdbInterfacesSorted = homologClusters.Select(a => pdbInterfaces.Where(b => a.Any(c => c.PdbId == b.PdbId && (char.ToUpperInvariant(c.ChainId[0]) == b.ChainId1 || char.ToUpperInvariant(c.ChainId[0]) == b.ChainId2))).ToList()).ToList();
var outputData = new List <string>();
foreach (var clusterIndex in homologClusterIndexes)
{
var cluster = pdbInterfacesSorted[clusterIndex - 1];
// currently, cluster is a list of chain1-->chain2 interfaces ... so the 'chain2' interface needs adding to the record
foreach (var inf1 in cluster)
{
var partner =
cluster.Where(a => a != inf1 && a.PdbId == inf1.PdbId && inf1.ChainId2 == a.ChainId1)
.OrderByDescending(
a => InterfaceOverlap(inf1.MinResSeq, inf1.MaxResSeq, a.MinResSeq, a.MaxResSeq))
.ToList();
var first = partner.FirstOrDefault();
if (first != null)
{
inf1.Partner1InterfaceAminoAcids = first.InterfaceAminoAcids;
inf1.Partner1InterfaceInteractionsMask = first.InterfaceInteractionsMask;
inf1.Partner1InterfaceOverlap = InterfaceOverlap(inf1.MinResSeq, inf1.MaxResSeq, first.MinResSeq, first.MaxResSeq);
}
var second = partner.ElementAtOrDefault(1);
if (second != null)
{
inf1.Partner2InterfaceAminoAcids = second.InterfaceAminoAcids;
inf1.Partner2InterfaceInteractionsMask = second.InterfaceInteractionsMask;
inf1.Partner2InterfaceOverlap = InterfaceOverlap(inf1.MinResSeq, inf1.MaxResSeq, second.MinResSeq, second.MaxResSeq);
}
}
cluster = cluster.Where(a => a.Partner1InterfaceOverlap > 0 || a.Partner2InterfaceOverlap > 0).ToList();
/*
* var partners =
* foreach (var inf2 in cluster)
* {
* if (inf1.PdbId!=inf2.PdbId) continue;
*
* if (inf1==inf2) continue;
*
* if (!(inf1.ChainId1==inf2.ChainId2 || inf1.ChainId2==inf2.ChainId1)) continue;
*
* //
* var overlap = InterfaceOverlap(inf1.MinResSeq, inf1.MaxResSeq, inf2.MinResSeq, inf2.MaxResSeq);
*
* if (overlap > 0)
* {
* if (overlap > inf1.Partner1InterfaceOverlap)
* {
* inf1.Partner1InterfaceOverlap = overlap;
* inf1.Partner1InterfaceAminoAcids = inf2.InterfaceAminoAcids;
* inf1.Partner1InterfaceInteractionsMask = inf2.InterfaceInteractionsMask;
* }
*
* if (overlap > inf2.Partner1InterfaceOverlap)
* {
* inf2.Partner1InterfaceOverlap = overlap;
* inf2.Partner1InterfaceAminoAcids = inf1.InterfaceAminoAcids;
* inf2.Partner1InterfaceInteractionsMask = inf1.InterfaceInteractionsMask;
* }
* }
* }
* }
*/
//var interfaces = cluster.Select(a => a.InterfaceAminoAcids).ToList();
//interfaces = interfaces.Where(a => interfaces.Count(b => b == a) > 1).ToList();
//cluster = cluster.Where(a => a.InterfaceAminoAcids.Length >= 5 && cluster.Count(b => b.InterfaceAminoAcids == a.InterfaceAminoAcids) > 1).ToList();
cluster = cluster.Where(a => a.InterfaceAminoAcids.Length >= 5).ToList();
var clusterInterfaces = cluster.Select(a => a.InterfaceAminoAcids).ToList();
var homologInterfaces = new List <List <string> >();
foreach (var inf1 in clusterInterfaces)
{
var highest_score = decimal.MinValue;
string highest_inf = null;
foreach (var inf2 in clusterInterfaces)
{
if (inf1 == inf2)
{
continue;
}
var score = FindAtomicContacts.AlignedSequenceSimilarityPercentage(inf1, inf2, FindAtomicContacts.AlignmentType.NMW);
if (score.Item1 > highest_score)
{
highest_score = score.Item1;
highest_inf = inf2;
}
}
var y = homologInterfaces.FirstOrDefault(a => a.Contains(inf1) || a.Contains(highest_inf));
if (y != null)
{
if (!y.Contains(inf1))
{
y.Add(inf1);
}
if (!y.Contains(highest_inf))
{
y.Add(highest_inf);
}
}
else
{
var z = new List <string>();
z.Add(inf1);
z.Add(highest_inf);
homologInterfaces.Add(z);
}
}
foreach (var c in cluster)
{
c.Partner1ClusterIndex = homologInterfaces.FindIndex(b => b.Contains(c.Partner1InterfaceAminoAcids));
c.Partner2ClusterIndex = homologInterfaces.FindIndex(b => b.Contains(c.Partner2InterfaceAminoAcids));
}
for (int index = 0; index < homologInterfaces.Count; index++)
{
var homologInterface = homologInterfaces[index];
var cluster2 =
cluster.Where(a => homologInterface.Contains(a.InterfaceAminoAcids)
)
.OrderBy(a => a.Partner1ClusterIndex)
.ThenBy(a => a.Partner2ClusterIndex)
.ThenBy(a => a.InterfaceAminoAcids)
.ThenBy(a => a.Partner1InterfaceAminoAcids)
.ThenBy(a => a.Partner2InterfaceAminoAcids)
.ToList();
var partners =
cluster2.Select(
a =>
new Tuple <string, string, string>(a.InterfaceAminoAcids, a.Partner1InterfaceAminoAcids,
a.Partner2InterfaceAminoAcids)).Distinct();
cluster2 =
partners.Select(
a =>
cluster2.FirstOrDefault(
b =>
b.InterfaceAminoAcids == a.Item1 && b.Partner1InterfaceAminoAcids == a.Item2 &&
b.Partner2InterfaceAminoAcids == a.Item3)).ToList();
outputData.Add("cluster " + clusterIndex + "." + index);
outputData.AddRange(cluster2.Select(a => a.ToString()).ToList());
outputData.Add("");
}
}
File.WriteAllLines(@"c:\ds96ub_homologs\ds96ub_homologs_interfaces.csv", outputData);//pdbInterfaces.Select(a=>a.ToString()).ToList());
}