public override long[] Generate(IAtomContainer container)
{
int[][] graph = ToAdjList(container);
Suppressed suppressed = suppression.Suppress(container);
return(Generate(seedGenerator.Generate(container), factory.Create(container, graph), graph, suppressed));
}
public void Bitset()
{
BitArray set = new BitArray(48);
set.Set(2, true);
set.Set(3, true);
set.Set(5, true);
set.Set(7, true);
set.Set(11, true);
set.Set(42, true);
Suppressed suppressed = Suppressed.FromBitSet(set);
Assert.IsTrue(suppressed.Contains(2));
Assert.IsTrue(suppressed.Contains(3));
Assert.IsTrue(suppressed.Contains(5));
Assert.IsTrue(suppressed.Contains(7));
Assert.IsTrue(suppressed.Contains(11));
Assert.IsTrue(suppressed.Contains(42));
Assert.IsFalse(suppressed.Contains(0));
Assert.IsFalse(suppressed.Contains(1));
Assert.IsFalse(suppressed.Contains(4));
Assert.IsFalse(suppressed.Contains(6));
Assert.IsFalse(suppressed.Contains(8));
Assert.IsFalse(suppressed.Contains(9));
Assert.IsFalse(suppressed.Contains(10));
Assert.IsFalse(suppressed.Contains(12));
Assert.IsFalse(suppressed.Contains(13));
Assert.IsFalse(suppressed.Contains(14));
}
public void None()
{
Suppressed suppressed = Suppressed.None;
for (int i = 0; i < 1000; i++)
{
Assert.IsFalse(suppressed.Contains(i));
}
}
public override Suppressed Suppress(IAtomContainer container)
{
BitArray hydrogens = new BitArray(container.Atoms.Count);
for (int i = 0; i < container.Atoms.Count; i++)
{
IAtom atom = container.Atoms[i];
hydrogens.Set(i, atom is IPseudoAtom);
}
return(Suppressed.FromBitSet(hydrogens));
}
public void TestGetUnsuppressed()
{
var suppression = AtomSuppression.Unsuppressed;
var m_container = new Mock <IAtomContainer>(); var container = m_container.Object;
Suppressed suppressed = suppression.Suppress(container);
Assert.IsFalse(suppressed.Contains(0));
Assert.IsFalse(suppressed.Contains(1));
Assert.IsFalse(suppressed.Contains(2));
Assert.IsFalse(suppressed.Contains(3));
Assert.IsFalse(suppressed.Contains(4));
}
public void TestRotation()
{
var m_seedMock = new Mock <IAtomHashGenerator>(); var seedMock = m_seedMock.Object;
var m_container = new Mock <IAtomContainer>(); var container = m_container.Object;
SuppressedAtomHashGenerator generator = new SuppressedAtomHashGenerator(new Mock <IAtomHashGenerator>().Object,
new Xorshift(), AtomSuppression.Unsuppressed, 2);
int[][] graph = new int[][] { new[] { 1, 2, 3 }, new[] { 0 }, new[] { 0 }, new[] { 0 } };
// simulate 3 identical neighbors
long[] invs = new long[] { 21, 31, 31, 31 };
long[] unique = new long[4];
long[] rotated = new long[4];
// non-suppressed
{
long value = generator.Next(graph, 0, invs, unique, rotated, Suppressed.None);
Assert.IsTrue(Compares.AreDeepEqual(new long[] { 31, 0, 0, 0 }, unique));
Assert.IsTrue(Compares.AreDeepEqual(new long[] { generator.Rotate(31, 2), 0, 0, 0 }, rotated));
Assert.IsTrue(Compares.AreDeepEqual(
value,
generator.Distribute(21) ^ 31 ^ generator.Rotate(31) ^ generator.Rotate(31, 2)));
}
// okay now suppress vertices 1
{
BitArray suppressed = new BitArray(3);
suppressed.Set(1, true);
long value = generator.Next(graph, 0, invs, unique, rotated, Suppressed.FromBitSet(suppressed));
Assert.IsTrue(Compares.AreDeepEqual(new long[] { 31, 0, 0, 0 }, unique));
Assert.IsTrue(Compares.AreDeepEqual(new long[] { generator.Rotate(31, 1), 0, 0, 0 }, rotated)); // 31 only encountered twice
Assert.IsTrue(Compares.AreDeepEqual(value, generator.Distribute(21) ^ 31 ^ generator.Rotate(31)));
}
// okay now suppress vertices 1 and 3
{
BitArray suppressed = new BitArray(4);
suppressed.Set(1, true);
suppressed.Set(3, true);
long value = generator.Next(graph, 0, invs, unique, rotated, Suppressed.FromBitSet(suppressed));
Assert.IsTrue(Compares.AreDeepEqual(new long[] { 31, 0, 0, 0 }, unique));
Assert.IsTrue(Compares.AreDeepEqual(new long[] { 31, 0, 0, 0 }, rotated)); // 31 only encountered once and is not rotated
Assert.IsTrue(Compares.AreDeepEqual(value, generator.Distribute(21) ^ 31)); // only encountered once
}
}
private long[] Generate(IAtomContainer container, long[] seeds, IStereoEncoder encoder, int[][] graph)
{
Suppressed suppressed = suppression.Suppress(container);
// compute original values then find indices equivalent values
long[] original = simple.Generate(seeds, encoder, graph, suppressed);
var equivalentSet = finder.Find(original, container, graph);
var equivalents = equivalentSet.ToArray();
// size of the matrix we need to make
int n = original.Length;
int m = equivalents.Length;
// skip when there are no equivalent atoms
if (m < 2)
{
return(original);
}
// matrix of perturbed values and identity values
long[][] perturbed = Arrays.CreateJagged <long>(n, m + 1);
// set the original values in the first column
for (int i = 0; i < n; i++)
{
perturbed[i][0] = original[i];
}
// systematically perturb equivalent vertex
for (int i = 0; i < m; i++)
{
int equivalentIndex = equivalents[i];
// perturb the value and reset stereo configuration
original[equivalentIndex] = Rotate(original[equivalentIndex]);
encoder.Reset();
// compute new hash codes and copy the values a column in the matrix
long[] tmp = simple.Generate(Copy(original), encoder, graph, suppressed);
for (int j = 0; j < n; j++)
{
perturbed[j][i + 1] = tmp[j];
}
// reset value
original[equivalentIndex] = perturbed[equivalentIndex][0];
}
return(Combine(perturbed));
}
public long[] Generate(IAtomContainer container)
{
Suppressed suppressed = suppression.Suppress(container);
int n = container.Atoms.Count;
int m = n - suppressed.Count; // number of non-suppressed vertices
int seed = m > 1 ? 9803 % m : 1;
long[] hashes = new long[n];
for (int i = 0; i < n; i++)
{
hashes[i] = Distribute(seed * encoder.Encode(container.Atoms[i], container));
}
return(hashes);
}
public void TestGenerate_Simple()
{
var m_seedMock = new Mock <IAtomHashGenerator>(); var seedMock = m_seedMock.Object;
var m_container = new Mock <IAtomContainer>(); var container = m_container.Object;
SuppressedAtomHashGenerator generator = new SuppressedAtomHashGenerator(new Mock <IAtomHashGenerator>().Object,
new Xorshift(), AtomSuppression.Unsuppressed, 2);
// no suppression
{
// first iteration, values are distributed and then neighbours xor'd
// in. when two neighbours have the same value the second should be
// rotated
long[] first = new long[] { generator.Distribute(1) ^ 2L,
generator.Distribute(2L) ^ 1L ^ generator.Rotate(1L), generator.Distribute(1) ^ 2L };
long[] second = new long[] { generator.Distribute(first[0]) ^ first[1],
generator.Distribute(first[1]) ^ first[0] ^ generator.Rotate(first[2]),
generator.Distribute(first[2]) ^ first[1] };
Assert.IsTrue(Compares.AreDeepEqual(
second,
generator.Generate(new long[] { 1L, 2L, 1L }, StereoEncoder.Empty, new int[][] { new[] { 1 }, new[] { 0, 2 }, new[] { 1 } },
Suppressed.None)));
}
// vertex '2' supressed
BitArray suppressed = new BitArray(3);
suppressed.Set(2, true);
{
long[] first = new long[] { generator.Distribute(1) ^ 2L, generator.Distribute(2L) ^ 1L, // generator.Rotate(1L) not included is '[2]' is suppressed
0L, };
long[] second = new long[] { generator.Distribute(first[0]) ^ first[1], // generator.Rotate(first[2]) not included is '[2]' is suppressed
generator.Distribute(first[1]) ^ first[0], 0L };
Assert.IsTrue(Compares.AreDeepEqual(
second,
generator.Generate(new long[] { 1L, 2L, 1L }, StereoEncoder.Empty, new int[][] { new[] { 1 }, new[] { 0, 2 }, new[] { 1 } },
Suppressed.FromBitSet(suppressed))));
}
}
public void TestGenerate_Disconnected()
{
var m_seedMock = new Mock <IAtomHashGenerator>(); var seedMock = m_seedMock.Object;
var m_container = new Mock <IAtomContainer>(); var container = m_container.Object;
SuppressedAtomHashGenerator generator = new SuppressedAtomHashGenerator(new Mock <IAtomHashGenerator>().Object,
new Xorshift(), AtomSuppression.Unsuppressed, 2);
// there are no neighbours, the values should be rotated
long expected = generator.Distribute(generator.Distribute(1));
Assert.IsTrue(Compares.AreDeepEqual(
new long[] { expected, expected, expected },
generator.Generate(new long[] { 1L, 1L, 1L }, StereoEncoder.Empty, new int[][] { new int[] { }, new int[] { }, new int[] { } },
Suppressed.None)));
BitArray suppressed = new BitArray(3);
suppressed.Set(1, true);
Assert.IsTrue(Compares.AreDeepEqual(
new long[] { expected, 0L, expected },
generator.Generate(new long[] { 1L, 1L, 1L }, StereoEncoder.Empty, new int[][] { new int[] { }, new int[] { }, new int[] { } },
Suppressed.FromBitSet(suppressed))));
}
public void TestGenerate_ZeroDepth()
{
var m_seedMock = new Mock <IAtomHashGenerator>(); var seedMock = m_seedMock.Object;
var m_container = new Mock <IAtomContainer>(); var container = m_container.Object;
SuppressedAtomHashGenerator generator = new SuppressedAtomHashGenerator(new Mock <IAtomHashGenerator>().Object,
new Xorshift(), AtomSuppression.Unsuppressed, 0);
Assert.IsTrue(Compares.AreDeepEqual(
new long[] { 1L, 1L, 1L },
generator.Generate(new long[] { 1L, 1L, 1L }, StereoEncoder.Empty, new int[][] { new int[] { }, new int[] { }, new int[] { } },
Suppressed.None)));
BitArray suppressed = new BitArray(3);
suppressed.Set(0, true);
suppressed.Set(2, true);
Assert.IsTrue(Compares.AreDeepEqual(
new long[] { 0L, 1L, 0L },
generator.Generate(new long[] { 1L, 1L, 1L }, StereoEncoder.Empty, new int[][] { new int[] { }, new int[] { }, new int[] { } },
Suppressed.FromBitSet(suppressed))));
}
public void TestAnyPseudos()
{
var suppression = AtomSuppression.AnyPseudos;
var m_container = new Mock <IAtomContainer>(); var container = m_container.Object;
m_container.SetupGet(n => n.Atoms.Count).Returns(5);
var m_carbon = new Mock <IAtom>(); var carbon = m_carbon.Object;
var m_pseudo = new Mock <IPseudoAtom>(); var pseudo = m_pseudo.Object;
m_container.SetupGet(n => n.Atoms[0]).Returns(carbon);
m_container.SetupGet(n => n.Atoms[1]).Returns(pseudo);
m_container.SetupGet(n => n.Atoms[2]).Returns(carbon);
m_container.SetupGet(n => n.Atoms[3]).Returns(carbon);
m_container.SetupGet(n => n.Atoms[4]).Returns(pseudo);
Suppressed suppressed = suppression.Suppress(container);
Assert.IsFalse(suppressed.Contains(0));
Assert.IsTrue(suppressed.Contains(1));
Assert.IsFalse(suppressed.Contains(2));
Assert.IsFalse(suppressed.Contains(3));
Assert.IsTrue(suppressed.Contains(4));
}
/// <summary>
/// Package-private method for generating the hash for the given molecule.
/// The initial invariants are passed as to the method along with an
/// adjacency list representation of the graph.
/// </summary>
/// <param name="current">initial invariants</param>
/// <param name="encoder"></param>
/// <param name="graph">adjacency list representation</param>
/// <param name="suppressed"></param>
/// <returns>hash codes for atoms</returns>
public override long[] Generate(long[] current, IStereoEncoder encoder, int[][] graph, Suppressed suppressed)
{
int n = graph.Length;
var next = Copy(current);
// buffers for including adjacent invariants
var unique = new long[n];
var included = new long[n];
while (encoder.Encode(current, next))
{
Copy(next, current);
}
for (int d = 0; d < depth; d++)
{
for (int v = 0; v < n; v++)
{
next[v] = Next(graph, v, current, unique, included);
}
Copy(next, current);
while (encoder.Encode(current, next))
{
Copy(next, current);
}
}
return(current);
}
/// <summary> /// Internal method invoked by 'molecule' hash generators. /// </summary> /// <param name="current">the current invariants</param> /// <param name="encoder">encoder used for encoding stereo-chemistry</param> /// <param name="graph">adjacency list representation of the molecule</param> /// <param name="suppressed">bit set marks vertices which are 'suppressed' (may be ignored)</param> /// <returns>the atom hash values</returns> public abstract long[] Generate(long[] current, IStereoEncoder encoder, int[][] graph, Suppressed suppressed);
/// <summary>
/// Determine the next value of the atom at index <paramref name="v"/>. The value is
/// calculated by combining the current values of adjacent atoms. When a
/// duplicate value is found it can not be directly included and is
/// <i>rotated</i> the number of times it has previously been seen.
/// </summary>
/// <param name="graph">adjacency list representation of connected atoms</param>
/// <param name="v">the atom to calculate the next value for</param>
/// <param name="current">the current values</param>
/// <param name="unique">buffer for working out which adjacent values are unique</param>
/// <param name="included">buffer for storing the rotated <i>unique</i> value, this value is <i>rotated</i> each time the same value is found.</param>
/// <param name="suppressed">bit set indicates which atoms are 'suppressed'</param>
/// <returns>the next value for <paramref name="v"/></returns>
internal long Next(int[][] graph, int v, long[] current, long[] unique, long[] included, Suppressed suppressed)
{
if (suppressed.Contains(v))
{
return(current[v]);
}
long invariant = Distribute(current[v]);
int nUnique = 0;
foreach (var w in graph[v])
{
// skip suppressed atom
if (suppressed.Contains(w))
{
continue;
}
long adjInv = current[w];
// find index of already included neighbor
int i = 0;
while (i < nUnique && unique[i] != adjInv)
{
++i;
}
// no match, then the value is unique, use adjInv
// match, then rotate the previously included value
included[i] = (i == nUnique) ? unique[nUnique++] = adjInv : Rotate(included[i]);
invariant ^= included[i];
}
return(invariant);
}
/// <summary>
/// Package-private method for generating the hash for the given molecule.
/// The initial invariants are passed as to the method along with an
/// adjacency list representation of the graph.
/// </summary>
/// <param name="current">initial invariants</param>
/// <param name="encoder"></param>
/// <param name="graph">adjacency list representation</param>
/// <param name="suppressed"></param>
/// <returns>hash codes for atoms</returns>
public override long[] Generate(long[] current, IStereoEncoder encoder, int[][] graph, Suppressed suppressed)
{
// for the stereo perception depending on how the
// (BasicPermutationParity) is done we need to set the value to be as
// high (or low) as possible
foreach (var i in suppressed.ToArray())
{
current[i] = long.MaxValue;
}
int n = graph.Length;
long[] next = Copy(current);
// buffers for including adjacent invariants
long[] unique = new long[n];
long[] included = new long[n];
while (encoder.Encode(current, next))
{
Copy(next, current);
}
for (int d = 0; d < depth; d++)
{
for (int v = 0; v < n; v++)
{
next[v] = Next(graph, v, current, unique, included, suppressed);
}
Copy(next, current);
while (encoder.Encode(current, next))
{
Copy(next, current);
}
}
// zero all suppressed values so they are not combined in any molecule
// hash
foreach (var i in suppressed.ToArray())
{
current[i] = 0L;
}
return(current);
}
