/// <summary>
/// Build contigs from graph. For contigs whose coverage is less than
/// the specified threshold, remove graph nodes belonging to them.
/// </summary>
/// <param name="deBruijnGraph">DeBruijn Graph.</param>
/// <param name="coverageThresholdForContigs">Coverage Threshold for contigs.</param>
/// <returns>Number of nodes removed.</returns>
public long RemoveLowCoverageContigs(DeBruijnGraph deBruijnGraph, double coverageThresholdForContigs)
{
if (deBruijnGraph == null)
{
throw new ArgumentNullException("deBruijnGraph");
}
if (coverageThresholdForContigs <= 0)
{
throw new ArgumentException("For removing low coverage contigs, coverage threshold should be a positive number");
}
_coverageThreshold = coverageThresholdForContigs;
_graph = deBruijnGraph;
DeBruijnGraph.ValidateGraph(deBruijnGraph);
ExcludeAmbiguousExtensions();
Parallel.ForEach(deBruijnGraph.GetNodes(), n => n.ComputeValidExtensions());
GetSimplePaths(false);
Parallel.ForEach(deBruijnGraph.GetNodes(),n=>n.UndoAmbiguousExtensions());
return deBruijnGraph.RemoveMarkedNodes();
}
/// <summary>
/// Delete nodes marked for erosion. Update adjacent nodes to update their extension tables.
/// After nodes are deleted, some new end-points might be created. We need to check for
/// dangling links at these new points. This list is returned in the out parameter.
///
/// TODO: Perhaps refactor code so that the graph is only manipulated by itself?
/// Might make it easier to implement future performance improvements, or cost performance
/// </summary>
/// <param name="graph">De Bruijn Graph.</param>
private static IList<DeBruijnNode> RemoveErodedNodes(DeBruijnGraph graph)
{
bool eroded = graph.RemoveMarkedNodes()>0;
IList<DeBruijnNode> graphNodes;
if (eroded)
{
graphNodes = graph.GetNodes().AsParallel().Where(n =>
{
bool wasEndPoint = (n.LeftExtensionNodesCount == 0 || n.RightExtensionNodesCount == 0);
n.RemoveMarkedExtensions();
// Check if this is a new end point.
return (wasEndPoint || (n.LeftExtensionNodesCount == 0 || n.RightExtensionNodesCount == 0));
}).ToList();
}
else
{
graphNodes = new List<DeBruijnNode>();
}
return graphNodes;
}