try lung with leiden

Author: quentinblampey

novae/data/dataset.py

@@ -110,15 +110,21 @@ def _init_dataset(self):
 
     def __len__(self) -> int:
         if self.sample_cells is not None:
-            return min(self.sample_cells, len(self.shuffled_obs_ilocs))
+            return min(self.sample_cells, self.n_obs)
 
         if self.training:
-            n_obs = len(self.shuffled_obs_ilocs)
-            return min(n_obs, max(Nums.MIN_DATASET_LENGTH, int(n_obs * Nums.MAX_DATASET_LENGTH_RATIO)))
+            return self._clipped_dataset_length()
 
         assert self.single_adata, "Multi-adata mode not supported for inference"
 
-        return len(self.obs_ilocs)
+        return self.n_obs
+
+    @property
+    def n_obs(self):
+        return sum(adata.n_obs for adata in self.adatas)
+
+    def _clipped_dataset_length(self) -> int:
+        return min(self.n_obs, max(Nums.MIN_DATASET_LENGTH, int(self.n_obs * Nums.MAX_DATASET_LENGTH_RATIO)))
 
     def __getitem__(self, index: int) -> dict[str, Data]:
         """Gets a sample from the dataset, with one "main" graph and its corresponding "view" graph (only during training).

scripts/revision/missing_domains.py

@@ -6,11 +6,11 @@
 import novae
 from novae._constants import Nums
 
-Nums.WARMUP_EPOCHS = 4
-Nums.WARMUP_ILOCS = 7
+Nums.WARMUP_EPOCHS = 3
+Nums.WARMUP_ILOCS = 6
 Nums.LEVEL_SUBSELECT = 10
 
-suffix = "_sub_select11"
+suffix = "_sub_select12"
 
 path = Path("/gpfs/workdir/blampeyq/novae/data/_lung_robustness")
 
@@ -41,7 +41,8 @@
 # model.fine_tune(adatas, min_prototypes_ratio=0.5, reference="largest")
 # model.compute_representations(adatas)
 
-obs_key = model.assign_domains(adatas, level=7)
+# obs_key = model.assign_domains(adatas, level=7)
+obs_key = model.assign_domains(adatas)
 
 model.plot_prototype_weights()
 plt.savefig(path / f"prototype_weights{suffix}.pdf", bbox_inches="tight")