public static List <string> RemoveStructuresWithIncorrectNumberOfChains(CancellationToken cancellationToken, string[] pdbFolders, List <string> pdbIdList = null, Dictionary <string, List <string> > pdbIdChainIdList = null, int numberChainsRequired = 2, ProgressActionSet progressActionSet = null, int totalThreads = -1)
{
if (pdbFolders == null || pdbFolders.Length == 0)
{
throw new ArgumentOutOfRangeException(nameof(pdbFolders));
}
if (pdbIdList == null || pdbIdList.Count == 0)
{
throw new ArgumentOutOfRangeException(nameof(pdbIdList));
}
if (progressActionSet == null)
{
throw new ArgumentNullException(nameof(progressActionSet));
}
var pdbFilesArray = ProteinDataBankFileOperations.GetPdbFilesArray(pdbFolders);
pdbFilesArray = ProteinDataBankFileOperations.RemoveNonWhiteListedPdbIdFromPdbFilesArray(pdbIdList, pdbFilesArray);
ProteinDataBankFileOperations.ShowMissingPdbFiles(pdbFilesArray, pdbIdList, progressActionSet);
WorkDivision <List <string> > workDivision = new WorkDivision <List <string> >(pdbFilesArray.Length, totalThreads);
ProgressActionSet.StartAction(pdbFilesArray.Length, progressActionSet);
for (int threadIndex = 0; threadIndex < workDivision.ThreadCount; threadIndex++)
{
int localThreadIndex = threadIndex;
Task <List <string> > task = Task.Run(() =>
{
var taskResult = new List <string>();
for (int pdbFileNumber = workDivision.ThreadFirstIndex[localThreadIndex]; pdbFileNumber <= workDivision.ThreadLastIndex[localThreadIndex]; pdbFileNumber++)
{
if (cancellationToken.IsCancellationRequested)
{
break;
}
try
{
string pdbFilename = pdbFilesArray[pdbFileNumber];
string proteinId = ProteinDataBankFileOperations.PdbIdFromPdbFilename(pdbFilename);
// Check if the file found is included in the white list.
if (pdbIdList != null && !pdbIdList.Contains(proteinId))
{
continue;
}
var sequenceChainIdList = pdbIdChainIdList != null ? (pdbIdChainIdList.ContainsKey(proteinId) ? pdbIdChainIdList[proteinId].ToArray() : null) : null;
int chainCount = ProteinDataBankFileOperations.PdbAtomicChainsCount(pdbFilename, sequenceChainIdList, numberChainsRequired);
if (chainCount != numberChainsRequired)
{
if (!taskResult.Contains(proteinId))
{
taskResult.Add(proteinId);
}
}
}
finally
{
workDivision.IncrementItemsCompleted(1);
ProgressActionSet.ProgressAction(1, progressActionSet);
ProgressActionSet.EstimatedTimeRemainingAction(workDivision.StartTicks, workDivision.ItemsCompleted, workDivision.ItemsToProcess, progressActionSet);
}
}
return(taskResult);
}, cancellationToken);
workDivision.TaskList.Add(task);
}
workDivision.WaitAllTasks();
ProgressActionSet.FinishAction(true, progressActionSet);
var result = new List <string>();
foreach (var task in workDivision.TaskList.Where(t => t != null && t.Result != null && t.IsCompleted && !t.IsCanceled && !t.IsFaulted))
{
result.AddRange(task.Result);
}
result = result.Distinct().ToList();
return(result);
}
/// <summary>
/// Generate stats of interactions... also removes proteins not meeting minimum interactions requirement.
/// </summary>
/// <param name="cancellationToken"></param>
/// <param name="pdbFolders"></param>
/// <param name="pdbIdList"></param>
/// <param name="progressBar"></param>
/// <param name="estimatedTimeRemainingLabel"></param>
/// <returns></returns>
public static Dictionary <string, decimal> CalculateStructureSymmetry(CancellationToken cancellationToken, decimal maxAtomInterationDistance, string[] pdbFolders, List <string> pdbIdList = null, Dictionary <string, List <string> > pdbIdChainIdList = null, ProgressActionSet progressActionSet = null, int totalThreads = -1)
{
if (pdbFolders == null || pdbFolders.Length == 0)
{
throw new ArgumentOutOfRangeException(nameof(pdbFolders));
}
if (pdbIdList == null || pdbIdList.Count == 0)
{
throw new ArgumentOutOfRangeException(nameof(pdbIdList));
}
if (progressActionSet == null)
{
throw new ArgumentOutOfRangeException(nameof(progressActionSet));
}
const int requiredNumberOfChains = 2;
string[] pdbFilesArray = ProteinDataBankFileOperations.RemoveNonWhiteListedPdbIdFromPdbFilesArray(pdbIdList, ProteinDataBankFileOperations.GetPdbFilesArray(pdbFolders));
WorkDivision <Dictionary <string, decimal> > workDivision = new WorkDivision <Dictionary <string, decimal> >(pdbFilesArray.Length, totalThreads);
ProteinDataBankFileOperations.ShowMissingPdbFiles(pdbFilesArray, pdbIdList, progressActionSet);
ProgressActionSet.StartAction(pdbFilesArray.Length, progressActionSet);
for (int threadIndex = 0; threadIndex < workDivision.ThreadCount; threadIndex++)
{
int localThreadIndex = threadIndex;
Task <Dictionary <string, decimal> > task = Task.Run(() =>
{
var taskResult = new Dictionary <string, decimal>();
for (int pdbFileNumber = workDivision.ThreadFirstIndex[localThreadIndex]; pdbFileNumber <= workDivision.ThreadLastIndex[localThreadIndex]; pdbFileNumber++)
{
if (cancellationToken.IsCancellationRequested)
{
break;
}
try
{
string pdbFilename = pdbFilesArray[pdbFileNumber];
string proteinId = ProteinDataBankFileOperations.PdbIdFromPdbFilename(pdbFilename);
// Check if the file found is included in the white list.
if (pdbIdList != null && !pdbIdList.Contains(proteinId))
{
continue;
}
var chainIdList = pdbIdChainIdList != null ? (proteinId != null && pdbIdChainIdList.ContainsKey(proteinId) ? pdbIdChainIdList[proteinId].ToArray() : null) : null;
// Get atom chains.
ProteinChainListContainer proteinFileChains = ProteinDataBankFileOperations.PdbAtomicChains(pdbFilename, chainIdList, requiredNumberOfChains, requiredNumberOfChains, true);
if (proteinFileChains == null || proteinFileChains.ChainList == null || proteinFileChains.ChainList.Count != 2 ||
proteinFileChains.ChainList[StaticValues.ChainA] == null || proteinFileChains.ChainList[StaticValues.ChainA].AtomList == null || proteinFileChains.ChainList[StaticValues.ChainA].AtomList.Count == 0 ||
proteinFileChains.ChainList[StaticValues.ChainB] == null || proteinFileChains.ChainList[StaticValues.ChainB].AtomList == null || proteinFileChains.ChainList[StaticValues.ChainB].AtomList.Count == 0)
{
continue;
}
// Make a list to save interactions found.
var interactionMatchPercentage = new InteractionMatchPercentage(proteinId);
List <AtomPair> interactions = SearchInteractions.FindInteractions(cancellationToken, maxAtomInterationDistance, pdbFilename, pdbIdChainIdList);
interactionMatchPercentage.IncrementTotalInteractions(interactions.Count);
for (int interactionsIndex = 0; interactionsIndex < interactions.Count; interactionsIndex++)
{
interactionMatchPercentage.AddResidueSequenceIndex(StaticValues.ChainA, interactions[interactionsIndex].Atom1.resSeq.FieldValue);
interactionMatchPercentage.AddResidueSequenceIndex(StaticValues.ChainB, interactions[interactionsIndex].Atom2.resSeq.FieldValue);
}
InteractionMatchPercentage.CalculatePercentageResult calculatedPercentage = interactionMatchPercentage.CalculatePercentage();
taskResult.Add(interactionMatchPercentage.ProteinId, calculatedPercentage.InteractionMatchPercentageAverage);
}
finally
{
workDivision.IncrementItemsCompleted(1);
ProgressActionSet.ProgressAction(1, progressActionSet);
ProgressActionSet.EstimatedTimeRemainingAction(workDivision.StartTicks, workDivision.ItemsCompleted, workDivision.ItemsToProcess, progressActionSet);
}
}
return(taskResult);
}, cancellationToken);
workDivision.TaskList.Add(task);
}
workDivision.WaitAllTasks();
var result = new Dictionary <string, decimal>();
foreach (var task in workDivision.TaskList.Where(t => t != null && t.Result != null && t.IsCompleted && !t.IsCanceled && !t.IsFaulted))
{
foreach (var kvp in task.Result)
{
//if (result.ContainsKey(kvp.Key))
//{
// Console.WriteLine("Key already exists: '" + kvp.Key + "'");
//}
result.Add(kvp.Key, kvp.Value);
}
}
return(result);
}