Merge pull request #22 from calico/snp-strand

SNP scoring optimizations
calico · Apr 9, 2024 · ea5231b · ea5231b
2 parents 9625573 + 95e0b09
commit ea5231b
Show file tree

Hide file tree

Showing 4 changed files with 134 additions and 304 deletions.
diff --git a/src/baskerville/dataset.py b/src/baskerville/dataset.py
@@ -319,7 +319,7 @@ def make_strand_transform(targets_df, targets_strand_df):
         targets_strand_df (pd.DataFrame): Targets DataFrame, with strand pairs collapsed.
 
     Returns:
-        scipy.sparse.csr_matrix: Sparse matrix to sum strand pairs.
+        scipy.sparse.dok_matrix: Sparse matrix to sum strand pairs.
     """
 
     # initialize sparse matrix
@@ -336,7 +336,6 @@ def make_strand_transform(targets_df, targets_strand_df):
             if target.identifier[-1] == "-":
                 sti += 1
         ti += 1
-    strand_transform = strand_transform.tocsr()
 
     return strand_transform
 

diff --git a/src/baskerville/scripts/hound_ism_bed.py b/src/baskerville/scripts/hound_ism_bed.py
@@ -276,7 +276,7 @@ def main():
                     alt_preds = np.array(alt_preds)
 
                     ism_scores = snps.compute_scores(
-                        ref_preds, alt_preds, options.snp_stats
+                        ref_preds, alt_preds, options.snp_stats, None
                     )
                     for snp_stat in options.snp_stats:
                         scores_h5[snp_stat][si, mi - mut_start, ni] = ism_scores[

diff --git a/src/baskerville/scripts/hound_ism_snp.py b/src/baskerville/scripts/hound_ism_snp.py
@@ -156,6 +156,7 @@ def main():
     else:
         targets_strand_df = targets_df
         strand_transform = None
+
     num_targets = targets_strand_df.shape[0]
 
     #################################################################
@@ -249,7 +250,7 @@ def main():
                     alt_preds = np.array(alt_preds)
 
                     ism_scores = snps.compute_scores(
-                        ref_preds, alt_preds, options.snp_stats
+                        ref_preds, alt_preds, options.snp_stats, None
                     )
                     for snp_stat in options.snp_stats:
                         scores_h5[snp_stat][si, mi - mut_start, ni] = ism_scores[