hmm: Resolve some, but not all, breaking API changes

etal · Sep 4, 2024 · e759acf · e759acf
1 parent ee6beab
commit e759acf
Showing 1 changed file with 31 additions and 25 deletions.
diff --git a/cnvlib/segmentation/hmm.py b/cnvlib/segmentation/hmm.py
@@ -95,25 +95,25 @@ def hmm_get_model(cnarr, method, diploid_parx_genome, processes):
     if method == "hmm-germline":
         state_names = ["loss", "neutral", "gain"]
         distributions = [
-            Normal(-1.0, stdev, frozen=True),
-            Normal(0.0, stdev, frozen=True),
-            Normal(0.585, stdev, frozen=True),
+            Normal([-1.0], [stdev], covariance_type="diag", frozen=True),
+            Normal([0.0], [stdev], covariance_type="diag", frozen=True),
+            Normal([0.585], [stdev], covariance_type="diag", frozen=True),
         ]
     elif method == "hmm-tumor":
         state_names = ["del", "loss", "neutral", "gain", "amp"]
         distributions = [
-            Normal(-2.0, stdev, frozen=False),
-            Normal(-0.5, stdev, frozen=False),
-            Normal(0.0, stdev, frozen=True),
-            Normal(0.3, stdev, frozen=False),
-            Normal(1.0, stdev, frozen=False),
+            Normal([-2.0], [stdev], covariance_type="diag", frozen=False),
+            Normal([-0.5], [stdev], covariance_type="diag", frozen=False),
+            Normal([0.0], [stdev], covariance_type="diag", frozen=True),
+            Normal([0.3], [stdev], covariance_type="diag", frozen=False),
+            Normal([1.0], [stdev], covariance_type="diag", frozen=False),
         ]
     else:
         state_names = ["loss", "neutral", "gain"]
         distributions = [
-            Normal(-1.0, stdev, frozen=False),
-            Normal(0.0, stdev, frozen=False),
-            Normal(0.585, stdev, frozen=False),
+            Normal([-1.0], [stdev], covariance_type="diag", frozen=False),
+            Normal([0.0], [stdev], covariance_type="diag", frozen=False),
+            Normal([0.585], [stdev], covariance_type="diag", frozen=False),
         ]
 
     n_states = len(distributions)
@@ -126,13 +126,16 @@ def hmm_get_model(cnarr, method, diploid_parx_genome, processes):
     transition_matrix = (
         np.identity(n_states) * 100 + np.ones((n_states, n_states)) / n_states
     )
-
-    model = DenseHMM.from_matrix(
-        transition_matrix,
-        distributions,
-        start_probabilities,
-        state_names=state_names,
-        name=method,
+    # Rescale so max is 1.0
+    transition_matrix /= transition_matrix.max()
+
+    model = DenseHMM(
+        distributions=distributions,
+        edges=transition_matrix,
+        starts=start_probabilities,
+        ends=start_probabilities,
+        #state_names=state_names,
+        #name=method,
     )
 
     model.fit(
@@ -172,8 +175,8 @@ def variants_in_segment(varr, segment, min_variants=50):
         observations = varr.mirrored_baf(above_half=True)
         state_names = ["neutral", "alt"]
         distributions = [
-            Normal(0.5, 0.1, frozen=True),
-            Normal(0.67, 0.1, frozen=True),
+            Normal([0.5], [0.1], covariance_type="diag", frozen=True),
+            Normal([0.67], [0.1], covariance_type="diag", frozen=True),
         ]
         n_states = len(distributions)
         # Starts -- prefer neutral
@@ -183,12 +186,15 @@ def variants_in_segment(varr, segment, min_variants=50):
         transition_matrix = (
             np.identity(n_states) * 100 + np.ones((n_states, n_states)) / n_states
         )
+        # Rescale so max is 1.0
+        transition_matrix /= transition_matrix.max()
         model = DenseHMM.from_matrix(
-            transition_matrix,
-            distributions,
-            start_probabilities,
-            state_names=state_names,
-            name="loh",
+            distributions=distributions,
+            edges=transition_matrix,
+            starts=start_probabilities,
+            ends=start_probabilities,
+            #state_names=state_names,
+            #name="loh",
         )
 
         model.fit(