mainline dev yolo

meteor/dsutils.py

@@ -96,55 +96,3 @@ def map_from_Fs(dataset, Fs, phis, map_res):
 
     return ccp4
 
-
-def from_gemmi(gemmi_mtz):
-
-    """
-    Construct DataSet from gemmi.Mtz object
-    
-    If the gemmi.Mtz object contains an M/ISYM column and contains duplicated
-    Miller indices, an unmerged DataSet will be constructed. The Miller indices
-    will be mapped to their observed values, and a partiality flag will be
-    extracted and stored as a boolean column with the label, ``PARTIAL``.
-    Otherwise, a merged DataSet will be constructed.
-    If columns are found with the ``MTZInt`` dtype and are labeled ``PARTIAL``
-    or ``CENTRIC``, these will be interpreted as boolean flags used to
-    label partial or centric reflections, respectively.
-    
-    Parameters
-    ----------
-    gemmi_mtz : gemmi.Mtz
-        gemmi Mtz object
-    
-    Returns
-    -------
-    rs.DataSet
-    """
-
-    dataset = rs.DataSet(spacegroup=gemmi_mtz.spacegroup, cell=gemmi_mtz.cell)
-
-    # Build up DataSet
-    for c in gemmi_mtz.columns:
-        dataset[c.label] = c.array
-        # Special case for CENTRIC and PARTIAL flags
-        if c.type == "I" and c.label in ["CENTRIC", "PARTIAL"]:
-            dataset[c.label] = dataset[c.label].astype(bool)
-        else:
-            dataset[c.label] = dataset[c.label].astype(c.type)
-    dataset.set_index(["H", "K", "L"], inplace=True)
-
-    # Handle unmerged DataSet. Raise ValueError if M/ISYM column is not unique
-    m_isym = dataset.get_m_isym_keys()
-    if m_isym and dataset.index.duplicated().any():
-        if len(m_isym) == 1:
-            dataset.merged = False
-            dataset.hkl_to_observed(m_isym[0], inplace=True)
-        else:
-            raise ValueError(
-                "Only a single M/ISYM column is supported for unmerged data"
-            )
-    else:
-        dataset.merged = True
-
-    return dataset
-

meteor/validate.py

@@ -1,26 +1,21 @@
 import numpy as np
 from scipy.stats import differential_entropy
 
-from . import mask
 
-
-def negentropy(X):
+def negentropy(samples: np.ndarray, tolerance: float = 1e-4) -> float:
     """
     Return negentropy (float) of X (numpy array)
-    """
-
-    # negetropy is the difference between the entropy of samples x
-    # and a Gaussian with same variance
-    # http://gregorygundersen.com/blog/2020/09/01/gaussian-entropy/
-
-    std = np.std(X)
-    # neg_e = np.log(std*np.sqrt(2*np.pi*np.exp(1))) - differential_entropy(X)
-    neg_e = 0.5 * np.log(2.0 * np.pi * std**2) + 0.5 - differential_entropy(X)
-    # assert neg_e >= 0.0 + 1e-8
-
-    return neg_e
-
 
+    The negetropy is defined as the difference between the entropy of a distribution
+    and a Gaussian with same variance.
 
+    citation:
+        http://gregorygundersen.com/blog/2020/09/01/gaussian-entropy/
+    """
 
+    std = np.std(samples.flatten())
+    neg_e = 0.5 * np.log(2.0 * np.pi * std**2) + 0.5 - differential_entropy(samples.flatten())
+    if not neg_e >= -tolerance:
+        raise ValueError(f"negentropy is a relatively big negative number {neg_e} that exceeds the tolerance {tolerance} -- something may have gone wrong")
 
+    return neg_e

pyproject.toml

@@ -7,6 +7,9 @@ authors = [
     { name = "Thomas Lane", email = "[email protected]" }
 ]
 dependencies = [
+    "numpy",
+    "scipy",
+    "skimage",
     "reciprocalspaceship",
 ]
 

test/unit/test_validate.py

@@ -0,0 +1,11 @@
+
+import numpy as np
+from meteor import validate
+
+def test_negentropy_gaussian() -> None:
+    n_samples = 100
+    samples = np.random.normal(size=n_samples)
+    negentropy = validate.negentropy(samples)
+
+    # negentropy should be small for a Gaussian sample
+    assert np.abs(negentropy) < 1e-5