Merge pull request #33 from aertslab/dev

small bug fixes
aertslab · Oct 2, 2024 · a08d805 · a08d805
2 parents 5d26e6b + 6e9e45b
commit a08d805
Show file tree

Hide file tree

Showing 5 changed files with 232 additions and 28 deletions.
diff --git a/docs/tutorials/model_training_and_eval.ipynb b/docs/tutorials/model_training_and_eval.ipynb
diff --git a/src/crested/_io.py b/src/crested/_io.py
@@ -475,7 +475,7 @@ def import_bigwigs(
     df = pd.DataFrame(
         data_matrix,
         columns=consensus_peaks["region"],
-        index=[os.path.basename(file).split(".")[0] for file in bw_files],
+        index=[os.path.basename(file).rpartition('.')[0].replace('.', '_') for file in bw_files],
     )
 
     # Create AnnData object

diff --git a/src/crested/pl/patterns/_modisco_results.py b/src/crested/pl/patterns/_modisco_results.py
@@ -248,13 +248,13 @@ def create_clustermap(
 
     data = pd.DataFrame(pattern_matrix)
 
-    if isinstance(color_palette, str):
-        palette = sns.color_palette(color_palette, len(set(classes)))
-    else:
-        palette = color_palette
+    #if isinstance(color_palette, str):
+    #    palette = sns.color_palette(color_palette, len(set(classes)))
+    #else:
+    #    palette = color_palette
 
-    class_lut = dict(zip(set(classes), palette))
-    row_colors = pd.Series(classes).map(class_lut)
+    #class_lut = dict(zip(set(classes), palette))
+    #row_colors = pd.Series(classes).map(class_lut)
 
     if pat_seqs is not None:
         plt.rc("text", usetex=False)  # Turn off LaTeX to speed up rendering

diff --git a/src/crested/tl/_modisco_utils.py b/src/crested/tl/_modisco_utils.py
@@ -368,19 +368,17 @@ def match_score_patterns(a: dict, b: dict) -> float:
     -------
     Match score between the patterns.
     """
-    #a_pos = a['pos_pattern']
-    #b_pos = b['pos_pattern']
 
-    #score = tangermeme_tomtom.tomtom(Qs = [_get_ic(a["contrib_scores"], a_pos).T], Ts = [_get_ic(b["contrib_scores"], b_pos).T])
     _,_,ic_a = compute_ic(a['ppm'])
     _,_,ic_b = compute_ic(b['ppm'])
-    #print(a['id'])
-    #vizsequence.plot_weights(ic_a)
-    #print(b['id'])
-    #vizsequence.plot_weights(ic_b)
-    score = tangermeme_tomtom.tomtom(Qs = [ic_a.T], Ts = [ic_b.T])
+    try:
+        score = tangermeme_tomtom.tomtom(Qs = [ic_a.T], Ts = [ic_b.T])[0,0][0]
+    except Exception as e:
+        print(f"Warning: TOMTOM error while comparing patterns {a['id']} and {b['id']}. Returning no match.")
+        print(f"Error details: {e}")
+        score = 1
 
-    log_score = -np.log10(max(score[0,0][0], 1e-12))
+    log_score = -np.log10(max(score, 1e-12))
 
     return log_score
 

diff --git a/src/crested/tl/_tfmodisco.py b/src/crested/tl/_tfmodisco.py
@@ -283,7 +283,7 @@ def match_to_patterns(
 
     if verbose:
         print(
-            f'Match between {pattern_id} and {all_patterns[str(match_idx)]["pattern"]["id"]} with similarity score {str(max_sim)}'
+            f'Match between {pattern_id} and {all_patterns[str(match_idx)]["pattern"]["id"]} with similarity score {max_sim:.2f}'
         )
 
     all_patterns[str(match_idx)]["instances"][pattern_id] = p