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(ProteinBioClass.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 =
ProteinBioClass.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 <Ds93UbInterface>();
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 Ds93UbInterface(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) == b.ChainId1 || char.ToUpperInvariant(c.ChainId) == 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 = ProteinBioClass.AlignedSequenceSimilarityPercentage(inf1, inf2, ProteinBioClass.AlignmentType.NMW);
if (score.Score > highest_score)
{
highest_score = score.Score;
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());
}
static void Main(string[] args)
{
// this program takes a fasta or pdb file and finds all matching homologs
// FindHomologs.exe "c:\ds96ub\ds96ub.fasta" * "c:\pdb\pdb_seqres.fasta" NMW Y 0.3 75 c:\pdb\
// alignment_type = (n)one, (s)imple, NMW, SWM
var query_sequence_file = args[0]; //query.fasta
var query_id_chain = args[1]; //1A2G:B
var target_sequence_file = args[2]; //targets.fasta
var alignment_type_str = args[3]; //NMW,SWM,SIM,NON
if (alignment_type_str == "*")
{
alignment_type_str = "NMW,SWM,SIM,NON";
}
var alignment_type_str_split = alignment_type_str.ToUpperInvariant().Split(new char[] { ',', ';', ' ', '\t' });
var compare_physicochemically = args[4]; //Y/N
var compare_physicochemically_bool = compare_physicochemically == "Y";
var min_similarity_str = args[5]; // 0.3
var max_len_difference = args[6];
var max_len_difference_int = int.Parse(max_len_difference);
var output_folder = args[7];
var minSimilarity = decimal.Parse(min_similarity_str);
var alignmentTypes = new List <ProteinBioClass.AlignmentType>();
if (alignment_type_str_split.Contains("NMW"))
{
alignmentTypes.Add(ProteinBioClass.AlignmentType.NMW);
}
if (alignment_type_str_split.Contains("SWM"))
{
alignmentTypes.Add(ProteinBioClass.AlignmentType.SWM);
}
if (alignment_type_str_split.Contains("SIM"))
{
alignmentTypes.Add(ProteinBioClass.AlignmentType.SIM);
}
if (alignment_type_str_split.Contains("NON") || alignmentTypes.Count == 0)
{
alignmentTypes.Add(ProteinBioClass.AlignmentType.NON);
}
if (alignmentTypes.Count < alignment_type_str_split.Length)
{
Console.WriteLine("; unknown alignment type");
return;
}
// load list of query sequences
var queryPdbid = query_id_chain.Split(new char[] { ':' })[0];
var queryChainid = (query_id_chain.Contains(":") ? query_id_chain.Split(new char[] { ':' })[1] : "*")[0];
var querySeq = Sequence.LoadSequenceFile(query_sequence_file, null);
var queryResults = querySeq.Where(a =>
{
var id = new ProteinBioClass.SequenceId(a.Id);
return((queryPdbid == "*" || id.PdbId.ToUpperInvariant() == queryPdbid.ToUpperInvariant()) &&
(queryChainid == '*' || id.ChainId == queryChainid));
}).ToList();
if (queryResults.Count == 0)
{
Console.WriteLine("; the query pdbids/chainids were not found");
return;
}
// load list of target sequences
var targetSeq = Sequence.LoadSequenceFile(target_sequence_file, new string[] { null, "", "protein" });
targetSeq = targetSeq.Where(a => a.Count() >= 50).ToList();
Console.WriteLine("; aligning " + queryResults.Count + " query sequences to " + targetSeq.Count + " target sequences");
// perform alignment
//var startTime = DateTime.Now;
//var progress = 0;
//var progressLock = new object();
//var tasks = new List<Task<StringBuilder>>();
var queryPdbIds = queryResults.Select(a => new ProteinBioClass.SequenceId(a.Id).PdbId);
var targetPdbIds = targetSeq.Select(a => new ProteinBioClass.SequenceId(a.Id).PdbId);
var queryPdbIdCounts = new Dictionary <string, int>();
foreach (var x in queryPdbIds)
{
if (!queryPdbIdCounts.ContainsKey(x))
{
queryPdbIdCounts.Add(x, 1);
}
else
{
queryPdbIdCounts[x]++;
}
}
var targetPdbIdCounts = new Dictionary <string, int>();
foreach (var x in targetPdbIds)
{
if (!targetPdbIdCounts.ContainsKey(x))
{
targetPdbIdCounts.Add(x, 1);
}
else
{
targetPdbIdCounts[x]++;
}
}
foreach (var _query in queryResults)
{
var _queryId = new ProteinBioClass.SequenceId(_query.Id);
var filename = (new DirectoryInfo(output_folder).FullName) + @"\homologs_" + _queryId.PdbId + _queryId.ChainId + @".csv";
// skip if already processed
if (File.Exists(filename) && new FileInfo(filename).Length > 0)
{
continue;
}
var totalQueryPdbIdChains = queryPdbIdCounts[_queryId.PdbId];
WorkDivision wd = new WorkDivision(targetSeq.Count);
for (var thread = 0; thread < wd.ThreadCount; thread++)
{
var query = _query;
var queryId = _queryId;
var lti = thread;
wd.TaskList.Add(Task.Run(() =>
{
var result = new List <HomologChain>();
for (var target = wd.ThreadFirstIndex[lti]; target <= wd.ThreadLastIndex[lti]; target++)
{
var targetobj = targetSeq[target];
if (max_len_difference_int != -1 && Math.Abs(targetobj.Count() - query.Count()) > max_len_difference_int)
{
continue;
}
var targetId = new ProteinBioClass.SequenceId(targetobj.Id);
//var totalTargetPdbIdChains = targetSeq.Count(a => FindAtomicContacts.SequenceIdToPdbIdAndChainId(a.ID).PdbId.ToUpperInvariant() == targetId.PdbId.ToUpperInvariant());
//var timeRemaining =
// TimeSpan.FromTicks(DateTime.Now.Subtract(startTime).Ticks *
// ((targetSeq.Count * queryResults.Count) - (progress + 1)) /
// (progress + 1));
foreach (var alignmentType in alignmentTypes)
{
var scores = ProteinBioClass.AlignedSequenceSimilarityPercentage(query, targetobj, alignmentType /*,
* compare_physicochemically_bool*/);
decimal percentSimilar;
if (compare_physicochemically_bool)
{
percentSimilar = scores.ScoreEvo;
}
else
{
percentSimilar = scores.Score;
}
if (percentSimilar >= minSimilarity)
{
result.Add(new HomologChain(
queryId.PdbId, queryId.ChainId, totalQueryPdbIdChains,
targetId.PdbId, targetId.ChainId, targetPdbIdCounts[targetId.PdbId],
alignmentType.ToString(),
scores.Score,
scores.ScoreEvo));
}
}
//if (progress % 20 == 0)
// Console.Write("\r{0}% eta {1} ",
// Math.Round((decimal)(progress + 1) / (decimal)(targetSeq.Count * queryResults.Count),
// 2)
// .ToString(CultureInfo.InvariantCulture),
// timeRemaining.ToString(@"d\d\:h\h\:m\m\:s\s",
// CultureInfo.InvariantCulture));
//lock (progressLock)
// progress++;
}
return(result);
}));
}
wd.WaitAllTasks();
var mergedlist = new List <string>();
mergedlist.Add("; " + _queryId.PdbId + ":" + _queryId.ChainId);
mergedlist.Add(String.Join(",",
new string[]
{
"query pdb id", "query chain id", "query chains",
"target pdb id", "target chain id", "target chains",
"alignment method", "sequence similarity", "sequence evo similarity"
}));
foreach (var t in wd.TaskList)
{
var tc = t as Task <List <HomologChain> >;
if (tc == null)
{
throw new Exception("task in tasklist was null");
}
mergedlist.AddRange(tc.Result.Select(a => a.ToString()).ToList());
}
if (string.IsNullOrWhiteSpace(output_folder))
{
Console.WriteLine(String.Join(Environment.NewLine, mergedlist));
}
else
{
File.WriteAllLines(filename, mergedlist);
}
}
}
public static List <SequenceIdentityClusterMember> ClusterSequenceByAlignedSequenceIdentity(List <Sequence> seqList, ProteinBioClass.AlignmentIdentityOption alignmentIdentityOption, decimal mininumClusterPairwiseSimilarity = 0.3m, decimal mininumEvoClusterPairwiseSimilarity = 0.3m)
{
var allsequences = seqList.Select(a => new Tuple <string, char, string>(new ProteinBioClass.SequenceId(a.Id).PdbId, new ProteinBioClass.SequenceId(a.Id).ChainId, Sequence.EscapeAminoAcidSequence(a.FullSequence))).ToList();
var sequences = allsequences.Select(a => a.Item3).Distinct().ToList();
var sequenceIds = sequences.Select(a => allsequences.Where(b => b.Item3 == a).ToList()).ToList();
var seqClusters = new List <List <string> >();
for (int x = 0; x < sequences.Count; x++)
{
var seq1 = sequences[x];
var newCluster = new List <string>();
newCluster.Add(seq1);
seqClusters.Add(newCluster);
}
for (int indexX = 0; indexX < sequences.Count; indexX++)
{
Console.WriteLine("Aligning sequence " + indexX);
var seqX = sequences[indexX];
//List<decimal> scoreList = new List<decimal>();
//List<decimal> scoreEvoList = new List<decimal>();
for (int indexY = 0; indexY < sequences.Count; indexY++)
{
if (indexY <= indexX)
{
continue;
}
var seqY = sequences[indexY];
if ((decimal)Math.Min(seqX.Length, seqY.Length) / (decimal)Math.Max(seqX.Length, seqY.Length) < mininumClusterPairwiseSimilarity)
{
continue;
}
var cluster1 = seqClusters.FirstOrDefault(a => a.Contains(seqX));
var cluster2 = seqClusters.FirstOrDefault(a => a.Contains(seqY));
if (cluster1 != null && cluster2 != null && cluster1 == cluster2)
{
continue;
}
var score = ProteinBioClass.AlignedSequenceSimilarityPercentage(seqX, seqY, ProteinBioClass.AlignmentType.NON, alignmentIdentityOption);
Console.WriteLine("1: " + seqX);
Console.WriteLine("2: " + seqY);
Console.WriteLine("Score1: " + score.Score);
Console.WriteLine("Score2: " + score.ScoreEvo);
if (score.Score < mininumClusterPairwiseSimilarity || score.ScoreEvo < mininumEvoClusterPairwiseSimilarity)
{
var x = ProteinBioClass.AlignedSequenceSimilarityPercentage(seqX, seqY, ProteinBioClass.AlignmentType.SIM, alignmentIdentityOption);
if (x.Score > score.Score)
{
score.Score = x.Score;
}
if (x.ScoreEvo > score.ScoreEvo)
{
score.ScoreEvo = x.ScoreEvo;
}
}
if (score.Score < mininumClusterPairwiseSimilarity || score.ScoreEvo < mininumEvoClusterPairwiseSimilarity)
{
var x = ProteinBioClass.AlignedSequenceSimilarityPercentage(seqX, seqY, ProteinBioClass.AlignmentType.NMW, alignmentIdentityOption);
if (x.Score > score.Score)
{
score = x;
}
if (x.ScoreEvo > score.ScoreEvo)
{
score.ScoreEvo = x.ScoreEvo;
}
}
if (score.Score >= mininumClusterPairwiseSimilarity && score.ScoreEvo >= mininumEvoClusterPairwiseSimilarity)
{
var newCluster = new List <string>();
newCluster.AddRange(cluster1);
newCluster.AddRange(cluster2);
seqClusters.Remove(cluster1);
seqClusters.Remove(cluster2);
seqClusters.Add(newCluster);
}
//scoreList.Add(score.Score);
//scoreEvoList.Add(score.ScoreEvo);
}
//Console.WriteLine("[" + string.Join(", ", scoreList.Select(a => String.Format("{0:0.00}", a)).ToList()) + "]");
//Console.WriteLine("[" + string.Join(", ", scoreEvoList.Select(a => String.Format("{0:0.00}", a)).ToList()) + "]");
}
seqClusters = seqClusters.OrderBy(a => a.Count).ToList();
var output = new List <SequenceIdentityClusterMember>();
for (var index = 0; index < seqClusters.Count; index++)
{
var seqCluster = seqClusters[index];
foreach (var item in seqCluster)
{
var indexIds = sequences.IndexOf(item);
var ids = sequenceIds[indexIds];
foreach (var id in ids)
{
output.Add(new SequenceIdentityClusterMember(index + 1, ProteinBioClass.PdbIdFromPdbFilename(id.Item1), id.Item2, id.Item3));
}
}
}
return(output);
}
private static void Main(string[] args)
{
// this program will load the homolog list in csv format and for homologs of X sequence distance return a list of all partners
// however, some partners may be duplicates, which cannot initially be removed, since they may bind differently in other instances
// then, because of such cases, unique id to describe each protein must be created... this is slightly problematic because
// close target homologs of proteins are also considered to be the same protein as the query protein
// which means that they could exist for more than one query protein
// FindHomologsCluster.exe c:\pdb\ds96ub_homologs\ c:\pdb\pdb_seqres.fasta 0.9 0.9 > ds96ub_homologs.csv
var homolog_csv_folder = args[0];
var sequence_file = args[1];
var min_similarity_str = args[2];
var min_similarity_evo_str = args[3];
var min_similarity = decimal.Parse(min_similarity_str);
var min_similarity_evo = decimal.Parse(min_similarity_evo_str);
var seqList = Sequence.LoadSequenceFile(sequence_file, new[] { null, "", "protein" });
var homologCsvFiles = Directory.GetFiles(homolog_csv_folder, "homologs_?????.csv");
var parsedData = FindHomologs.FindHomologs.HomologChain.Load(homologCsvFiles);
Array.Clear(homologCsvFiles, 0, homologCsvFiles.Length);
//var query_pdb_list = parsed_data.Select(a => new Tuple<string, string>(a.query_pdb_id, a.query_chainid)).ToList();
//var target_pdb_list = parsed_data.Select(a => new Tuple<string, string>(a.target_pdb_id, a.target_chainid)).ToList();
//var query_alignments = new List<homolog_csv>();
var homologs_clustered = new List <List <Tuple <string, char> > >();
//var min_similarity = 0.9m;
foreach (var rec in parsedData)
{
if (rec.AlignmentScore >= min_similarity && rec.AlignmentScoreEvo >= min_similarity_evo)
{
//var query_group = homologs_clustered.FirstOrDefault(a => a.FirstOrDefault(b => b.Item1 == rec.query_pdb_id && b.Item2 == rec.query_chainid) != null);
//var target_group = homologs_clustered.FirstOrDefault(a => a.FirstOrDefault(b => b.Item1 == rec.target_pdb_id && b.Item2 == rec.target_chainid) != null);
List <Tuple <string, char> > query_group = null;
List <Tuple <string, char> > target_group = null;
foreach (var cluster in homologs_clustered)
{
var xq = cluster.FirstOrDefault(b => b.Item1.ToUpperInvariant() == rec.QueryPdbId.ToUpperInvariant() && b.Item2 == rec.QueryChainId);
if (xq == null)
{
continue;
}
query_group = cluster;
break;
}
foreach (var cluster in homologs_clustered)
{
var xt =
cluster.FirstOrDefault(b => b.Item1.ToUpperInvariant() == rec.TargetPdbId.ToUpperInvariant() && b.Item2 == rec.TargetChainId);
if (xt == null)
{
continue;
}
target_group = cluster;
break;
}
var new_group = new List <Tuple <string, char> >();
if (query_group != null)
{
new_group.AddRange(query_group);
homologs_clustered.Remove(query_group);
query_group.Clear();
}
else
{
new_group.Add(new Tuple <string, char>(rec.QueryPdbId, rec.QueryChainId));
}
if (target_group != null)
{
new_group.AddRange(target_group);
homologs_clustered.Remove(target_group);
target_group.Clear();
}
else
{
new_group.Add(new Tuple <string, char>(rec.TargetPdbId, rec.TargetChainId));
}
new_group = new_group.Distinct().ToList(); // try without distinct?
new_group = new_group.OrderBy(a => a.Item1).ThenBy(a => a.Item2).ToList();
homologs_clustered.Add(new_group);
}
}
var seq_list_ids = seqList.Select(a => new ProteinBioClass.SequenceId(a.Id)).ToList();
var wd2 = new WorkDivision(homologs_clustered.Count);
for (var thread2 = 0; thread2 < wd2.ThreadCount; thread2++)
{
var lti2 = thread2;
wd2.TaskList.Add(Task.Run(() =>
{
var result2 = new List <string>();
for (var index2 = wd2.ThreadFirstIndex[lti2]; index2 <= wd2.ThreadLastIndex[lti2]; index2++)
{
var cluster2 = homologs_clustered[index2];
var wd3 = new WorkDivision(cluster2.Count);
for (var thread3 = 0; thread3 < wd3.ThreadCount; thread3++)
{
var lti3 = thread3;
var cluster3 = cluster2;
var index4 = index2;
wd3.TaskList.Add(Task.Run(() =>
{
var result = new List <HomologClusterData>();
for (var index3 = wd3.ThreadFirstIndex[lti3]; index3 <= wd3.ThreadLastIndex[lti3]; index3++)
{
var item = cluster3[index3];
Sequence s = null;
for (var j = 0; j < seqList.Count; j++)
{
if (seq_list_ids[j].PdbId.ToUpperInvariant() == item.Item1.ToUpperInvariant() && seq_list_ids[j].ChainId == item.Item2)
{
s = seqList[j];
break;
}
}
if (s == null)
{
throw new Exception("sequence not found for " + item.Item1 + ":" + item.Item2);
}
var complexChains = seq_list_ids.Count(a => a.PdbId.ToUpperInvariant() == item.Item1.ToUpperInvariant());
var minAlignmentScore = -1m;
var maxAlignmentScore = -1m;
var minAlignmentScoreEvo = -1m;
var maxAlignmentScoreEvo = -1m;
foreach (var item2 in cluster3)
{
if (ReferenceEquals(item, item2))
{
continue;
}
Sequence s2 = null;
for (var j2 = 0; j2 < seqList.Count; j2++)
{
if (seq_list_ids[j2].PdbId.ToUpperInvariant() == item2.Item1.ToUpperInvariant() &&
seq_list_ids[j2].ChainId == item2.Item2)
{
s2 = seqList[j2];
break;
}
}
if (s2 == null)
{
continue;
}
var alignmentScore = ProteinBioClass.AlignedSequenceSimilarityPercentage(s,
s2,
ProteinBioClass.AlignmentType.NMW);
if (alignmentScore.Score > maxAlignmentScore || maxAlignmentScore == -1m)
{
maxAlignmentScore = alignmentScore.Score;
}
if (alignmentScore.Score < minAlignmentScore || minAlignmentScore == -1m)
{
minAlignmentScore = alignmentScore.Score;
}
if (alignmentScore.ScoreEvo > maxAlignmentScoreEvo || maxAlignmentScoreEvo == -1m)
{
maxAlignmentScoreEvo = alignmentScore.ScoreEvo;
}
if (alignmentScore.ScoreEvo < minAlignmentScoreEvo || minAlignmentScoreEvo == -1m)
{
minAlignmentScoreEvo = alignmentScore.ScoreEvo;
}
}
var r = new HomologClusterData(index4 + 1, index3 + 1, item.Item1, item.Item2, complexChains, Convert.ToInt32(s.Count()), minAlignmentScore, maxAlignmentScore, minAlignmentScoreEvo, maxAlignmentScoreEvo, s.FullSequence);
result.Add(r);
}
return(result);
}));
}
wd3.WaitAllTasks();
result2.Add("; Cluster # " + (index2 + 1) + " with " + wd3.ItemsToProcess + " protein chains");
result2.Add("cluster index,item index,pdb id,chain id,complex chains,seq len,min clstr sid,max clstr sid,min evo clstr sid,max evo clstr sid,sequence");
foreach (var task in wd3.TaskList)
{
//if (task.IsFaulted || task.IsCanceled) continue;
var tr = task as Task <List <HomologClusterData> >;
if (tr == null || tr.Result == null)
{
continue;
}
result2.AddRange(tr.Result.Select(a => a.ToString()).ToList());
}
result2.Add("");
}
return(result2);
}));
//wd2.TaskList.Add(task2);
}
wd2.WaitAllTasks();
var result1 = new List <string>();
foreach (var task in wd2.TaskList)
{
//if (task.IsFaulted || task.IsCanceled) continue;
var tr = task as Task <List <string> >;
if (tr == null || tr.Result == null)
{
continue;
}
result1.AddRange(tr.Result);
}
foreach (var line in result1)
{
Console.WriteLine(line);
}
// partners may have other interfaces, should those also be considered?
}
