Remove gambit.sigs.convert module

src/gambit/sigs/calc.py

@@ -278,3 +278,48 @@ def calc_file_signatures(kspec: KmerSpec,
 		assert all(sig is not None for sig in sigs)
 
 	return SignatureList(sigs, kspec)
+
+
+def dense_to_sparse(vec: Sequence[bool]) -> KmerSignature:
+	"""Convert k-mer set from dense bit vector to sparse coordinate representation.
+
+	Parameters
+	----------
+	vec
+		Boolean vector indicating which k-mers are present.
+
+	Returns
+	-------
+	numpy.ndarray
+		Sorted array  of coordinates of k-mers present in vector. Data type will be ``numpy.intp``.
+
+	See Also
+	--------
+	.sparse_to_dense
+	"""
+	return np.flatnonzero(vec)
+
+
+def sparse_to_dense(k_or_kspec: Union[int, KmerSpec],  coords: KmerSignature) -> np.ndarray:
+	"""Convert k-mer set from sparse coordinate representation back to dense bit vector.
+
+	Parameters
+	----------
+	k_or_kspec
+		Value of k or a :class:`.KmerSpec` instance.
+	coords
+		Sparse coordinate array.
+
+	Returns
+	-------
+	numpy.ndarray
+		Dense k-mer bit vector.
+
+	See Also
+	--------
+	.dense_to_sparse
+	"""
+	idx_len = k_or_kspec.nkmers if isinstance(k_or_kspec, KmerSpec) else nkmers(k_or_kspec)
+	vec = np.zeros(idx_len, dtype=np.bool_)
+	vec[coords] = 1
+	return vec

tests/common.py

@@ -7,7 +7,7 @@
 from gambit.kmers import KmerSpec, kmer_to_index
 from gambit.seq import seq_to_bytes, revcomp
 from gambit.sigs import KmerSignature, SignatureArray
-from gambit.sigs.convert import dense_to_sparse, sparse_to_dense
+from gambit.sigs.calc import dense_to_sparse, sparse_to_dense
 from gambit.db import Taxon
 
 

tests/sigs/test_calc.py

@@ -7,7 +7,8 @@
 from Bio import SeqIO
 from Bio.Seq import Seq
 
-from gambit.sigs.calc import calc_signature, calc_file_signature, calc_file_signatures
+from gambit.sigs.calc import calc_signature, calc_file_signature, calc_file_signatures, \
+	dense_to_sparse, sparse_to_dense
 from gambit.kmers import KmerSpec, index_to_kmer
 from gambit.seq import SEQ_TYPES, revcomp, SequenceFile
 import gambit.util.io as ioutil
@@ -171,3 +172,28 @@ def test_calc_file_signatures(self, record_sets, files, concurrency):
 			sigs2 = calc_file_signatures(KSPEC, files, progress=pconf, concurrency=concurrency)
 
 		assert sigarray_eq(sigs, sigs2)
+
+
+def test_dense_sparse_conversion():
+	"""Test conversion between dense and sparse representations of k-mer coordinates."""
+
+	for k in range(1, 10):
+
+		kspec = KmerSpec(k, 'ATGAC')
+
+		# Create dense signature with every 3rd k-mer
+		vec = np.zeros(kspec.nkmers, dtype=bool)
+		vec[np.arange(vec.size) % 3 == 0] = True
+
+		# Convert to sparse
+		sig = dense_to_sparse(vec)
+
+		assert len(sig) == vec.sum()
+		for index in sig:
+			assert vec[index]
+
+		# Check sorted
+		assert np.all(np.diff(sig) > 0)
+
+		# Check converting back
+		assert np.array_equal(vec, sparse_to_dense(kspec, sig))

tests/test_metric.py

@@ -7,7 +7,7 @@
 
 from gambit.metric import jaccard, jaccarddist, jaccard_bits, jaccard_generic, jaccarddist_array, \
 	jaccarddist_matrix, jaccarddist_pairwise, num_pairs, SCORE_DTYPE, BOUNDS_DTYPE
-from gambit.sigs.convert import sparse_to_dense
+from gambit.sigs.calc import sparse_to_dense
 from gambit.sigs import SignatureArray, SignatureList, dump_signatures, load_signatures
 from gambit.kmers import KmerSpec
 from gambit.util.progress import check_progress

tests/test_tests_common.py

@@ -5,7 +5,7 @@
 
 from gambit.kmers import KmerSpec, kmer_to_index, nkmers
 from gambit.seq import revcomp
-from gambit.sigs.convert import dense_to_sparse
+from gambit.sigs.calc import dense_to_sparse
 from gambit.util.progress import get_progress
 from . import common