Functional refactor

.gitignore

@@ -4,6 +4,7 @@ wandb/
 *slurm*
 tests/data/genomes
 tests/data/test_pipeline
+tests/data/test_contribution_scores/
 
 # Sphinx documentation
 _build
@@ -12,4 +13,4 @@ node_modules
 .DS_Store
 ._.DS_Store
 docs/tutorials/mouse_biccn.ipynb
-docs/tutorials/.ipynb_checkpoints
+docs/tutorials/.ipynb_checkpoints

docs/api/utils.md

@@ -17,5 +17,6 @@ CREsted provides a few utility function to help with sequence encoding, function
     fetch_sequences
     reverse_complement
     permute_model
+    calculate_gc_distribution
     setup_logging
 ```

src/crested/_genome.py

@@ -151,7 +151,14 @@ def name(self) -> str:
                 return basename
         return self._name
 
-    def fetch(self, chrom=None, start=None, end=None, strand="+", region=None) -> str:
+    def fetch(
+        self,
+        chrom: str | None = None,
+        start: int | None = None,
+        end: int | None = None,
+        strand: str = "+",
+        region: str | None = None,
+    ) -> str:
         """
         Fetch a sequence from a genomic region.
 
@@ -196,6 +203,13 @@ def fetch(self, chrom=None, start=None, end=None, strand="+", region=None) -> st
         else:
             return seq
 
+    def __repr__(self) -> str:
+        """Return a string representation of the Genome object."""
+        fasta_exists = self.fasta is not None
+        chrom_sizes_exists = self.chrom_sizes is not None
+        annotations_exists = self.annotation is not None
+        return f"Genome({self.name}, fasta={fasta_exists}, chrom_sizes={chrom_sizes_exists}, annotation={annotations_exists})"
+
 
 def register_genome(genome: Genome):
     """

src/crested/tl/__init__.py

@@ -7,6 +7,13 @@
 from . import data, losses, metrics, zoo
 from ._configs import TaskConfig, default_configs
 from ._crested import Crested
+from ._tools import (
+    contribution_scores,
+    contribution_scores_specific,
+    extract_layer_embeddings,
+    predict,
+    score_gene_locus,
+)
 
 if find_spec("modiscolite") is not None:
     MODISCOLITE_AVAILABLE = True
@@ -35,6 +42,11 @@
     "TaskConfig",
     "default_configs",
     "Crested",
+    "extract_layer_embeddings",
+    "predict",
+    "contribution_scores",
+    "contribution_scores_specific",
+    "score_gene_locus",
 ]
 
 if MODISCOLITE_AVAILABLE:

src/crested/tl/_crested.py

@@ -324,9 +324,6 @@ def fit(
         if self.model and (
             not hasattr(self.model, "optimizer") or self.model.optimizer is None
         ):
-            logger.warning(
-                "Model does not have an optimizer. Please compile the model before training."
-            )
             self.model.compile(
                 optimizer=self.config.optimizer,
                 loss=self.config.loss,