updating test and data class

pinellolab · Nov 25, 2023 · 2804ecc · 2804ecc
1 parent 526490f
commit 2804ecc
Show file tree

Hide file tree

Showing 2 changed files with 51 additions and 12 deletions.
diff --git a/bean/preprocessing/data_class.py b/bean/preprocessing/data_class.py
@@ -929,7 +929,7 @@ def __init__(
         replicate_column: str = "rep",
         **kwargs,
     ):
-        self._pre_init(condition_column)
+        screen = self._pre_init(screen, condition_column, time_column, condition_column)
         super().__init__(
             screen=screen,
             repguide_mask=repguide_mask,
@@ -942,8 +942,13 @@ def __init__(
         )
         self._post_init()
 
-    def _pre_init(self, time_column: str, condition_column: str):
-        self.time_column = time_column
+    def _pre_init(
+        self,
+        time_column: str,
+        condition_column: str,
+        control_condition: str,
+        normalized_time_column: str = "_time_normed",
+    ):
         if not np.issubdtype(self.screen.samples[time_column].dtype, np.number):
             raise ValueError(
                 f"Invalid timepoint value({self.screen.samples[time_column]}) in screen.samples[{time_column}]: check input."
@@ -957,6 +962,28 @@ def _pre_init(self, time_column: str, condition_column: str):
                 f"Not all replicate share same timepoint definition. If you have missing bin data, add the sample and add 'mask' column in 'screen.samples', or run `bean-qc` that automatically handles this. \n{self.screen.samples}"
             )
 
+        self.screen.samples[normalized_time_column] = (
+            self.screen.samples[time_column] - self.screen.samples[time_column].min()
+        ) / (
+            self.screen.samples[time_column].max()
+            - self.screen.samples[time_column].min()
+        )
+        self.screen_selected.samples[normalized_time_column] = (
+            self.screen_selected.samples[time_column]
+            - self.screen.samples[time_column].min()
+        ) / (
+            self.screen.samples[time_column].max()
+            - self.screen.samples[time_column].min()
+        )
+        self.screen_control.samples[normalized_time_column] = (
+            self.screen_control.samples[time_column]
+            - self.screen.samples[time_column].min()
+        ) / (
+            self.screen.samples[time_column].max()
+            - self.screen.samples[time_column].min()
+        )
+        self.time_column = normalized_time_column
+
     def _post_init(
         self,
     ):
@@ -1212,6 +1239,7 @@ def __init__(
         replicate_column="rep",
         condition_column="condition",
         time_column="time",
+        control_condition="bulk",
         control_can_be_selected=True,
         target_col="target",
         sample_mask_column="mask",
@@ -1226,12 +1254,15 @@ def __init__(
             sample_mask_column=sample_mask_column,
             replicate_column=replicate_column,
             condition_column=condition_column,
+            control_condition=control_condition,
             time_column=time_column,
             shrink_alpha=shrink_alpha,
             control_can_be_selected=control_can_be_selected,
             **kwargs,
         )
-        SurvivalScreenData._pre_init(self, time_column, condition_column)
+        SurvivalScreenData._pre_init(
+            self, time_column, condition_column, control_condition
+        )
         ScreenData._post_init(self)
         SurvivalScreenData._post_init(self)
         VariantScreenData._post_init(self, target_col)
@@ -1282,6 +1313,7 @@ def __init__(
         replicate_column="rep",
         condition_column="condition",
         time_column="time",
+        control_condition="control",
         control_can_be_selected=True,
         target_col="target",
         sample_mask_column="mask",
@@ -1299,12 +1331,15 @@ def __init__(
             sample_mask_column=sample_mask_column,
             replicate_column=replicate_column,
             condition_column=condition_column,
+            control_condition=control_condition,
             time_column=time_column,
             shrink_alpha=shrink_alpha,
             control_can_be_selected=control_can_be_selected,
             **kwargs,
         )
-        SurvivalScreenData._pre_init(self, time_column, condition_column)
+        SurvivalScreenData._pre_init(
+            self, time_column, condition_column, control_condition
+        )
         ScreenData._post_init(self)
         SurvivalScreenData._post_init(self)
         VariantScreenData._post_init(self, target_col)
@@ -1328,6 +1363,7 @@ def __init__(
         replicate_column="rep",
         condition_column="condition",
         time_column="time",
+        control_condition="control",
         control_can_be_selected=True,
         sample_mask_column="mask",
         use_const_pi: bool = False,
@@ -1347,12 +1383,15 @@ def __init__(
             sample_mask_column=sample_mask_column,
             replicate_column=replicate_column,
             condition_column=condition_column,
+            control_condition=control_condition,
             time_column=time_column,
             shrink_alpha=shrink_alpha,
             control_can_be_selected=control_can_be_selected,
             **kwargs,
         )
-        SurvivalScreenData._pre_init(self, time_column, condition_column)
+        SurvivalScreenData._pre_init(
+            self, time_column, condition_column, control_condition
+        )
         ScreenData._post_init(self)
         SurvivalScreenData._post_init(self)
         TilingReporterScreenData._post_init(

diff --git a/tests/test_run.py b/tests/test_run.py
@@ -4,7 +4,7 @@
 
 @pytest.mark.order(13)
 def test_run_variant_wacc():
-    cmd = "bean-run variant tests/data/var_mini_screen_annotated.h5ad --scale-by-acc --acc-bw-path tests/data/accessibility_signal_chr6.bw -o tests/test_res/var/ --repguide-mask None"
+    cmd = "bean-run sorting variant tests/data/var_mini_screen_annotated.h5ad --scale-by-acc --acc-bw-path tests/data/accessibility_signal_chr6.bw -o tests/test_res/var/ --repguide-mask None"
     try:
         subprocess.check_output(
             cmd,
@@ -17,7 +17,7 @@ def test_run_variant_wacc():
 
 @pytest.mark.order(14)
 def test_run_variant_noacc():
-    cmd = "bean-run variant tests/data/var_mini_screen_annotated.h5ad -o tests/test_res/var/ "
+    cmd = "bean-run sorting variant tests/data/var_mini_screen_annotated.h5ad -o tests/test_res/var/ "
     try:
         subprocess.check_output(
             cmd,
@@ -30,7 +30,7 @@ def test_run_variant_noacc():
 
 @pytest.mark.order(15)
 def test_run_variant_wo_negctrl_uniform():
-    cmd = "bean-run variant tests/data/var_mini_screen_annotated.h5ad -o tests/test_res/var/ --uniform-edit "
+    cmd = "bean-run sorting variant tests/data/var_mini_screen_annotated.h5ad -o tests/test_res/var/ --uniform-edit "
     try:
         subprocess.check_output(
             cmd,
@@ -43,7 +43,7 @@ def test_run_variant_wo_negctrl_uniform():
 
 @pytest.mark.order(16)
 def test_run_tiling_wo_negctrl():
-    cmd = "bean-run tiling tests/data/tiling_mini_screen_annotated.h5ad --scale-by-acc --acc-bw-path tests/data/accessibility_signal.bw -o tests/test_res/tiling/ --allele-df-key allele_counts_spacer_0_19_A.G_translated_prop0.1_0.3 --control-guide-tag None  --repguide-mask None"
+    cmd = "bean-run sorting tiling tests/data/tiling_mini_screen_annotated.h5ad --scale-by-acc --acc-bw-path tests/data/accessibility_signal.bw -o tests/test_res/tiling/ --allele-df-key allele_counts_spacer_0_19_A.G_translated_prop0.1_0.3 --control-guide-tag None  --repguide-mask None"
     try:
         subprocess.check_output(
             cmd,
@@ -56,7 +56,7 @@ def test_run_tiling_wo_negctrl():
 
 @pytest.mark.order(17)
 def test_run_tiling_with_wo_negctrl_noacc():
-    cmd = "bean-run tiling tests/data/tiling_mini_screen_annotated.h5ad -o tests/test_res/tiling/ --allele-df-key allele_counts_spacer_0_19_A.G_translated_prop0.1_0.3 --control-guide-tag None  --repguide-mask None"
+    cmd = "bean-run sorting tiling tests/data/tiling_mini_screen_annotated.h5ad -o tests/test_res/tiling/ --allele-df-key allele_counts_spacer_0_19_A.G_translated_prop0.1_0.3 --control-guide-tag None  --repguide-mask None"
     try:
         subprocess.check_output(
             cmd,
@@ -69,7 +69,7 @@ def test_run_tiling_with_wo_negctrl_noacc():
 
 @pytest.mark.order(18)
 def test_run_tiling_with_wo_negctrl_uniform():
-    cmd = "bean-run tiling tests/data/tiling_mini_screen_annotated.h5ad -o tests/test_res/tiling/ --uniform-edit --allele-df-key allele_counts_spacer_0_19_A.G_translated_prop0.1_0.3 --control-guide-tag None  --repguide-mask None"
+    cmd = "bean-run sorting tiling tests/data/tiling_mini_screen_annotated.h5ad -o tests/test_res/tiling/ --uniform-edit --allele-df-key allele_counts_spacer_0_19_A.G_translated_prop0.1_0.3 --control-guide-tag None  --repguide-mask None"
     try:
         subprocess.check_output(
             cmd,