diff --git a/pyproject.toml b/pyproject.toml
index 8e7fcd940..0c37a1431 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -13,6 +13,7 @@ authors = [
 dependencies = [
     "ase",
     "torch",
+    "rascaline-torch @ git+https://github.com/luthaf/rascaline#subdirectory=python/rascaline-torch",
     "metatensor-core",
     "metatensor-torch"
 ]
@@ -54,9 +55,7 @@ requires = [
 build-backend = "setuptools.build_meta"
 
 [project.optional-dependencies]
-soap-bpnn = [
-    "rascaline-torch @ git+https://github.com/luthaf/rascaline#subdirectory=python/rascaline-torch",
-]
+soap-bpnn = []
 
 [tool.setuptools.packages.find]
 where = ["src"]
diff --git a/src/metatensor_models/readers/__init__.py b/src/metatensor_models/readers/__init__.py
new file mode 100644
index 000000000..27441dfcb
--- /dev/null
+++ b/src/metatensor_models/readers/__init__.py
@@ -0,0 +1,44 @@
+""""Readers for structures and target values."""
+
+from typing import List, Dict, Optional
+
+from pathlib import Path
+
+from metatensor.torch import TensorMap
+
+from .structures import STRUCTURE_READERS
+from .targets import TARGET_READERS
+
+from rascaline.torch.system import Systems
+
+
+def read_structures(filename: str, fileformat: Optional[str] = None) -> List[Systems]:
+    """Reads a structure information from file."""
+
+    if fileformat is None:
+        fileformat = Path(filename).suffix
+
+    try:
+        reader = STRUCTURE_READERS[fileformat]
+    except KeyError:
+        raise ValueError(f"fileformat '{fileformat}' is not supported")
+
+    return reader(filename)
+
+
+def read_targets(
+    filename: str,
+    target_value: str,
+    fileformat: Optional[str] = None,
+) -> Dict[str, TensorMap]:
+    """Reads target information from file."""
+
+    if fileformat is None:
+        fileformat = Path(filename).suffix
+
+    try:
+        reader = TARGET_READERS[fileformat]
+    except KeyError:
+        raise ValueError(f"fileformat '{fileformat}' is not supported")
+
+    return reader(filename, target_value)
diff --git a/src/metatensor_models/readers/structures/__init__.py b/src/metatensor_models/readers/structures/__init__.py
new file mode 100644
index 000000000..123a893df
--- /dev/null
+++ b/src/metatensor_models/readers/structures/__init__.py
@@ -0,0 +1,3 @@
+from .ase import read_ase
+
+STRUCTURE_READERS = {".xyz": read_ase}
diff --git a/src/metatensor_models/readers/structures/ase.py b/src/metatensor_models/readers/structures/ase.py
new file mode 100644
index 000000000..26d94ba35
--- /dev/null
+++ b/src/metatensor_models/readers/structures/ase.py
@@ -0,0 +1,11 @@
+from typing import List
+
+import ase.io
+from rascaline.systems import AseSystem
+from rascaline.torch.system import Systems, systems_to_torch
+
+
+def read_ase(filename: str) -> List[Systems]:
+    systems = [AseSystem(atoms) for atoms in ase.io.read(filename, ":")]
+
+    return systems_to_torch(systems)
diff --git a/src/metatensor_models/readers/targets/__init__.py b/src/metatensor_models/readers/targets/__init__.py
new file mode 100644
index 000000000..69a98d1a8
--- /dev/null
+++ b/src/metatensor_models/readers/targets/__init__.py
@@ -0,0 +1,3 @@
+from .ase import read_ase
+
+TARGET_READERS = {".xyz": read_ase}
diff --git a/src/metatensor_models/readers/targets/ase.py b/src/metatensor_models/readers/targets/ase.py
new file mode 100644
index 000000000..160535b18
--- /dev/null
+++ b/src/metatensor_models/readers/targets/ase.py
@@ -0,0 +1,38 @@
+from typing import Dict, List, Union
+
+import ase.io
+import torch
+from metatensor.torch import Labels, TensorBlock, TensorMap
+
+
+def read_ase(
+    filename: str,
+    target_values: Union[List[str], str],
+) -> Dict[str, TensorMap]:
+    """Store target informations from file in a :class:`metatensor.TensorMap`.
+
+    :returns:
+        TensorMap containing the given information
+    """
+
+    if type(target_values) is str:
+        target_values = [target_values]
+
+    frames = ase.io.read(filename, ":")
+
+    target_dictionary = {}
+    for target_value in target_values:
+        values = [f.info[target_value] for f in frames]
+
+        n_structures = len(values)
+
+        block = TensorBlock(
+            values=torch.tensor(values).reshape(-1, 1),
+            samples=Labels(["structure"], torch.arange(n_structures).reshape(-1, 1)),
+            components=[],
+            properties=Labels([target_value], torch.tensor([(0,)])),
+        )
+
+        target_dictionary[target_value] = TensorMap(Labels.single(), [block])
+
+    return target_dictionary