[REF] abagen.io functions don't copy dataframes by default

abagen/allen.py

@@ -344,7 +344,7 @@ def get_expression_data(atlas, atlas_info=None, *, exact=True,
         files.update(dict(microarray=micro, pacall=micro, annotation=annot))
 
     # get dataframe of probe information (reannotated or otherwise)
-    probe_info = io.read_probes(files['probes'][0])
+    probe_info = io.read_probes(files['probes'][0], copy=True)
     if reannotated:
         lgr.info('Reannotating microarray probes with information from '
                  'Arnatkevic̆iūtė et al., 2018, NeuroImage')

abagen/io.py

@@ -36,6 +36,7 @@ def _make_parquet(fname, convert_only=False):
     data : pandas.DataFrame
         Data loaded from `fname`
     """
+
     # get ideal parquet filename
     parqname = fname.rpartition('.csv')[0] + '.parq'
 
@@ -55,7 +56,7 @@ def _make_parquet(fname, convert_only=False):
     return data
 
 
-def read_microarray(fname, parquet=True):
+def read_microarray(fname, copy=False, parquet=True):
     """
     Loads MicroarrayExpression.csv file found at `fname`
 
@@ -66,6 +67,9 @@ def read_microarray(fname, parquet=True):
     ----------
     fname : str
         Path to MicroarrayExpression.csv file
+    copy : bool, optional
+        Whether to return a copy if `fname` is a pre-loaded pandas.Dataframe.
+        Default: False
     parquet : bool, optional
         Whether to load data from parquet file instead of CSV. If a parquet
         file does not already exist then one will be created for faster loading
@@ -79,28 +83,29 @@ def read_microarray(fname, parquet=True):
         and `S` is samples. The row index is the unique probe ID assigned
         during processing, which can be used to match data to the information
         obtained with :func:`read_probes`. The column index is the unique
-        sample ID (integer, beginning at 1) which can be used to match data to
+        sample ID (integer, beginning at 0) which can be used to match data to
         the information obtained with :func:`read_annotation`.
     """
-    if not isinstance(fname, str):
-        if isinstance(fname, pd.DataFrame):
-            return fname.copy()
-        else:
-            raise TypeError('Provided fname {} must be a filepath.'
-                            .format(fname))
 
-    if use_parq and parquet:
-        data = _make_parquet(fname, convert_only=False).set_index('0')
-    else:
-        data = pd.read_csv(fname, header=None, index_col=0)
-
-    data.index.name = 'probe_id'
-    data.columns = pd.Series(range(len(data.columns)), name='sample_id')
+    try:
+        if use_parq and parquet:
+            data = _make_parquet(fname, convert_only=False)
+            data = data.set_index('0')
+        else:
+            data = pd.read_csv(fname, header=None, index_col=0)
+        data.index.name = 'probe_id'
+        data.columns = pd.Series(range(len(data.columns)), name='sample_id')
+    except (AttributeError, ValueError):
+        if not isinstance(fname, pd.DataFrame):
+            raise TypeError('Provided fname must be filepath to Microarray'
+                            'Expression.csv file from Allen Human Brain '
+                            'Atlas.')
+        data = fname.copy() if copy else fname
 
     return data
 
 
-def read_ontology(fname, parquet=True):
+def read_ontology(fname, copy=False):
     """
     Loads Ontology.csv file found at `fname`
 
@@ -116,8 +121,9 @@ def read_ontology(fname, parquet=True):
     ----------
     fname : str
         Path to Ontology.csv file
-    parquet : bool, optional
-        Does nothing; for compatibility with other :mod:`abagen.io` functions
+    copy : bool, optional
+        Whether to return a copy if `fname` is a pre-loaded pandas.Dataframe.
+        Default: False
 
     Returns
     -------
@@ -127,17 +133,19 @@ def read_ontology(fname, parquet=True):
         'parent_structure_id', 'hemisphere', 'graph_order',
         'structure_id_path', and 'color_hex_triplet'.
     """
-    if not isinstance(fname, str):
-        if isinstance(fname, pd.DataFrame):
-            return fname.copy()
-        else:
-            raise TypeError('Provided fname {} must be a filepath.'
-                            .format(fname))
 
-    return pd.read_csv(fname)
+    try:
+        data = pd.read_csv(fname)
+    except ValueError:
+        if not isinstance(fname, pd.DataFrame):
+            raise TypeError('Provided fname must be filepath to Ontology.csv'
+                            'file from Allen Human Brain Atlas.')
+        data = fname.copy() if copy else fname
 
+    return data
 
-def read_pacall(fname, parquet=True):
+
+def read_pacall(fname, copy=False, parquet=True):
     """
     Loads PACall.csv file found at `fname`
 
@@ -158,6 +166,9 @@ def read_pacall(fname, parquet=True):
     ----------
     fname : str
         Path to PACall.csv file
+    copy : bool, optional
+        Whether to return a copy if `fname` is a pre-loaded pandas.Dataframe.
+        Default: False
     parquet : bool, optional
         Whether to load data from parquet file instead of CSV. If a parquet
         file does not already exist then one will be created for faster loading
@@ -175,25 +186,25 @@ def read_pacall(fname, parquet=True):
         the unique sample ID (integer, beginning at 1) which can be used to
         match data to the information obtained with :func:`read_annotation`.
     """
-    if not isinstance(fname, str):
-        if isinstance(fname, pd.DataFrame):
-            return fname.copy()
-        else:
-            raise TypeError('Provided fname {} must be a filepath.'
-                            .format(fname))
 
-    if use_parq and parquet:
-        data = _make_parquet(fname, convert_only=False).set_index('0')
-    else:
-        data = pd.read_csv(fname, header=None, index_col=0)
-
-    data.index.name = 'probe_id'
-    data.columns = pd.Series(range(len(data.columns)), name='sample_id')
+    try:
+        if use_parq and parquet:
+            data = _make_parquet(fname, convert_only=False)
+            data = data.set_index('0')
+        else:
+            data = pd.read_csv(fname, header=None, index_col=0)
+        data.index.name = 'probe_id'
+        data.columns = pd.Series(range(len(data.columns)), name='sample_id')
+    except (AttributeError, ValueError):
+        if not isinstance(fname, pd.DataFrame):
+            raise TypeError('Provided fname must be filepath to PACall.csv'
+                            'file from Allen Human Brain Atlas.')
+        data = fname.copy() if copy else fname
 
     return data
 
 
-def read_probes(fname, parquet=True):
+def read_probes(fname, copy=False):
     """
     Loads Probes.csv file found at `fname`
 
@@ -207,8 +218,9 @@ def read_probes(fname, parquet=True):
     ----------
     fname : str
         Path to Probes.csv file
-    parquet : bool, optional
-        Does nothing; for compatibility with other :mod:`abagen.io` functions
+    copy : bool, optional
+        Whether to return a copy if `fname` is a pre-loaded pandas.Dataframe.
+        Default: False
 
     Returns
     -------
@@ -219,17 +231,19 @@ def read_probes(fname, parquet=True):
         :func:`read_pacall`. Columns include 'probe_name', 'gene_id',
         'gene_symbol', 'gene_name', 'entrez_id', and 'chromosome'.
     """
-    if not isinstance(fname, str):
-        if isinstance(fname, pd.DataFrame):
-            return fname.copy()
-        else:
-            raise TypeError('Provided fname {} must be a filepath.'
-                            .format(fname))
 
-    return pd.read_csv(fname, index_col=0)
+    try:
+        data = pd.read_csv(fname, index_col=0)
+    except ValueError:
+        if not isinstance(fname, pd.DataFrame):
+            raise TypeError('Provided fname must be filepath to Probes.csv'
+                            'file from Allen Human Brain Atlas.')
+        data = fname.copy() if copy else fname
+
+    return data
 
 
-def read_annotation(fname, parquet=True):
+def read_annotation(fname, copy=False):
     """
     Loads SampleAnnot.csv file found at `fname`
 
@@ -243,8 +257,9 @@ def read_annotation(fname, parquet=True):
     ----------
     fname : str
         Path to SampleAnnot.csv file
-    parquet : bool, optional
-        Does nothing; for compatibility with other :mod:`abagen.io` functions
+    copy : bool, optional
+        Whether to return a copy if `fname` is a pre-loaded pandas.Dataframe.
+        Default: False
 
     Returns
     -------
@@ -257,14 +272,14 @@ def read_annotation(fname, parquet=True):
         'structure_acronym', 'structure_name', 'polygon_id', 'mri_voxel_x',
         'mri_voxel_y', 'mri_voxel_z', 'mni_x', 'mni_y', 'mni_z'.
     """
-    if not isinstance(fname, str):
-        if isinstance(fname, pd.DataFrame):
-            return fname.copy()
-        else:
-            raise TypeError('Provided fname {} must be a filepath.'
-                            .format(fname))
 
-    annotation = pd.read_csv(fname)
-    annotation.index.name = 'sample_id'
+    try:
+        data = pd.read_csv(fname)
+        data.index.name = 'sample_id'
+    except ValueError:
+        if not isinstance(fname, pd.DataFrame):
+            raise TypeError('Provided fname must be filepath to Annotation.'
+                            '.csv file from Allen Human Brain Atlas.')
+        data = fname.copy() if copy else fname
 
-    return annotation
+    return data

abagen/probes.py

@@ -45,11 +45,12 @@ def reannotate_probes(probes):
     reannot = resource_filename('abagen', 'data/reannotated.csv.gz')
     with gzip.open(reannot, 'r') as src:
         reannot = pd.read_csv(StringIO(src.read().decode('utf-8')))
-    reannot = reannot[['probe_name', 'gene_symbol', 'entrez_id']]
 
     # merge reannotated with original, keeping only reannotated
-    probes = io.read_probes(probes).reset_index()[['probe_name', 'probe_id']]
-    merged = pd.merge(reannot, probes, on='probe_name', how='left')
+    probes = io.read_probes(probes).reset_index()
+    merged = pd.merge(reannot[['probe_name', 'gene_symbol', 'entrez_id']],
+                      probes[['probe_name', 'probe_id']],
+                      on='probe_name', how='left')
 
     # reset index as probe_id and sort
     reannotated = merged.set_index('probe_id').sort_index()
@@ -88,7 +89,7 @@ def filter_probes(pacall, probes, threshold=0.5):
 
     threshold = np.clip(threshold, 0.0, 1.0)
 
-    probes = io.read_probes(probes)
+    probes = io.read_probes(probes, copy=True)
     signal, samples = [], 0
     for fname in pacall:
         data = io.read_pacall(fname).loc[probes.index]
@@ -581,6 +582,8 @@ def collapse_probes(microarray, annotation, probes, method='diff_stability'):
 
     # read in microarray data for all subjects; this can be quite slow...
     probes = io.read_probes(probes)
-    exp = [io.read_microarray(micro).loc[probes.index] for micro in microarray]
+    exp = [
+        io.read_microarray(m, copy=True).loc[probes.index] for m in microarray
+    ]
 
     return [e.T for e in collfunc(exp, probes, annotation)]

abagen/samples.py

@@ -85,7 +85,7 @@ def update_mni_coords(annotation):
                                         axis=1)
     coords = coords.set_index('well_id')
 
-    annotation = io.read_annotation(annotation)
+    annotation = io.read_annotation(annotation, copy=True)
 
     # basic check that all well_ids in annotation are present in coords
     # a future pandas update may cause this to raise a KeyError but we want
@@ -436,9 +436,9 @@ def _mirror_samples(microarray, pacall, annotation, ontology):
         Loaded input data with all samples duplicated across hemispheres
     """
 
-    microarray = io.read_microarray(microarray)
-    pacall = io.read_pacall(pacall)
-    annotation = io.read_annotation(annotation)
+    microarray = io.read_microarray(microarray, copy=True)
+    pacall = io.read_pacall(pacall, copy=True)
+    annotation = io.read_annotation(annotation, copy=True)
     ontology = io.read_ontology(ontology)
 
     # take all lh and rh samples and flip x-coordinate
@@ -480,7 +480,7 @@ def _mirror_ontology(annotation, ontology):
 
     HEMI_SWAP = dict(L='R', R='L')
 
-    annotation = io.read_annotation(annotation)
+    annotation = io.read_annotation(annotation, copy=True)
     ontology = io.read_ontology(ontology)
 
     # structure IDs are specific to structure + hemisphere, so we can use this