add template for gpu support

alntools/base.py

@@ -58,6 +58,13 @@ def __init__(self, enh: bool = False,
  self.sigma_factor = sigma_factor
  self.gap_penalty = gap_penalty
 
+ # TODO implement this
+ def submatrix_to_span(self, densitymap, mode: str = 'results') -> pd.DataFrame:
+ '''
+ run plmblast flow from substitution matrix
+ func should return same results as embedding_to_span
+ '''
+ pass
 
  def embedding_to_span(self, X: np.ndarray, Y: np.ndarray, mode: str = 'results' ) -> pd.DataFrame:
  '''

examples/scope.sh

@@ -1,12 +1,12 @@
 #!/bin/bash
 
 ALIGNMENT_CUTOFF="0.3"
-COSINE_CUTOFF=95
+COSINE_CUTOFF=70
 
 DBDIR=/home/nfs/kkaminski/PLMBLST/db/scop40-201
 RESULTS="/home/nfs/kkaminski/PLMBLST/results/scope40-201_${ALIGNMENT_CUTOFF}_${COSINE_CUTOFF}.csv"
 
-NUM_WORKERS=10
+NUM_WORKERS=6
 # Return hits with scores >=0.3
 # no cutoff because our database is small we dont need additional filtering
 # Run plm-blast

tests/test_utils.py

@@ -0,0 +1,19 @@
+import pytest
+
+
+
+
+@pytest.mark.parametrize("size", [10, 50, 100])
+def test_remove_duplicates(size):
+ '''
+ test if function properly filter redundant hits eg. a-b, a-b, a-c func should exclude b-a
+ '''
+
+ data = {idx: list(range(size)) for idx in range(size)}
+ # func
+
+ # expected number of samples is size*(size - 1)/2
+ expected_size = size*(size - 1)/2
+ actual_size = sum([len(v) for v in data.values()])
+
+ assert actual_size == expected_size