def test_compress_sample_metadata_missing_sample_from_sid2idx(self):
# And if sid2idx is missing samples that are in the metadata, that's
# also not allowed!
# Subset the sample metadata to remove Sample3
diff_sid2idx = {"Sample1": 0, "Sample2": 1}
with self.assertRaisesRegex(
ValueError,
"The sample IDs in the metadata's index and s_ids_to_indices are "
"not identical."):
compress_sample_metadata(diff_sid2idx, self.sm_ef)
def test_compress_sample_metadata_missing_sample_from_metadata(self):
# If the metadata is missing samples described in sid2idx, that's bad!
# ...And also probably impossible, unless someone messes up the code :P
# Subset the sample metadata to remove Sample1
diff_sm = self.sm_ef.copy()
diff_sm = diff_sm.drop(labels="Sample1", axis="index")
with self.assertRaisesRegex(
ValueError,
"The sample IDs in the metadata's index and s_ids_to_indices are "
"not identical."):
compress_sample_metadata(self.sid2idx, diff_sm)
def test_compress_sample_metadata_nonstr_vals(self):
# Test the "basic" case, just looking at our default data.
sm_copy = self.sm_ef.copy()
sid2idx_copy = deepcopy(self.sid2idx)
sm_cols, sm_vals = compress_sample_metadata(sid2idx_copy, sm_copy)
# As with compress_table(), verify that the inputs were left untouched.
assert_frame_equal(sm_copy, self.sm_ef)
self.assertEqual(sid2idx_copy, self.sid2idx)
self.assertEqual(sm_cols,
["Metadata1", "Metadata2", "Metadata3", "Metadata4"])
# For ease-of-reading, here's the metadata from above:
# "Metadata1": [0, 0, 0],
# "Metadata2": [0, 0, 0],
# "Metadata3": [1, 2, 3],
# "Metadata4": ["abc", "def", "ghi"]
# Check that the metadata values were all converted to strings and are
# structured properly.
self.assertEqual(
sm_vals,
[
["0", "0", "1", "abc"], # Sample1's metadata
["0", "0", "2", "def"], # Sample2's metadata
["0", "0", "3", "ghi"] # Sample3's metadata
])
def test_compress_sample_metadata_nonstr_columns(self):
diff_sm = self.sm_ef.copy()
diff_sm.columns = [100, 200, 'asdf', 2.5]
sm_cols, sm_vals = compress_sample_metadata(self.sid2idx, diff_sm)
# Main thing: check that the columns were converted to strings
self.assertEqual(sm_cols, ["100", "200", "asdf", "2.5"])
# (Also check that this didn't mess up the values for some reason)
self.assertEqual(
sm_vals,
[
["0", "0", "1", "abc"], # Sample1's metadata
["0", "0", "2", "def"], # Sample2's metadata
["0", "0", "3", "ghi"] # Sample3's metadata
])
def verify_fails_due_to_sid2idx(sid2idx):
with self.assertRaisesRegex(
ValueError,
r"Indices \(values\) of s_ids_to_indices are invalid."):
compress_sample_metadata(sid2idx, self.sm_ef)
def to_dict(self):
"""Convert processed data into a dictionary
Warning: the object returned by to_dict will contain references to
internal variables. Exercise caution if modifying the value of objects
returned by to_dict.
Returns
-------
dict
A dictionary describing the plots contained in the ordination
object and the sample + feature metadata.
"""
s_ids = f_ids = cmp_table = sm_cols = compressed_sm = None
sid2idxs = fid2idxs = {}
if self.is_community_plot:
# The fid2idxs dict we get from compress_table() is temporary --
# later, we'll restructure it so that the keys (feature IDs) are
# nodes' postorder positions in the tree rather than arbitrary
# unique integers. (TODO: it should be possible to speed this up by
# passing the tree to compress_table() so postorder positions can
# immediately be used as keys / feature IDs without an intermediate
# step.)
s_ids, f_ids, sid2idxs, fid2idxs_t, cmp_table = compress_table(
self.table)
sm_cols, compressed_sm = compress_sample_metadata(
sid2idxs, self.samples)
fm_cols, compressed_tm_tmp, compressed_im_tmp = \
compress_feature_metadata(self.tip_md, self.int_md)
# Use nodes' postorder positions as their "IDs" for the BIOM table and
# feature metadata
compressed_tm = {}
compressed_im = {}
# bptree indices start at one, hence we pad the arrays
names = [-1]
lengths = [-1]
for i in range(1, len(self.tree) + 1):
node = self.tree.postorderselect(i)
name = self.tree.name(node)
names.append(name)
lengths.append(self.tree.length(node))
if self.is_community_plot and name in fid2idxs_t:
fid2idxs[i] = fid2idxs_t[name]
f_ids[fid2idxs[i]] = i
if name in compressed_tm_tmp:
compressed_tm[i] = compressed_tm_tmp[name]
# Note: for internal metadata, node names may not be unique. Thus,
# we duplicate the internal node metadata for each node in the
# metadata with the same name.
if name in compressed_im_tmp:
compressed_im[i] = compressed_im_tmp[name]
data_to_render = {
'base_url': self.base_url,
# tree info
'tree': shifting(self.tree.B),
'lengths': lengths,
'names': names,
# Should we show sample metadata coloring / animation panels?
'is_community_plot': self.is_community_plot,
# Are we working with an EMPire plot?
'is_empire_plot': self.is_empire_plot,
# feature table
's_ids': s_ids,
'f_ids': f_ids,
's_ids_to_indices': sid2idxs,
'f_ids_to_indices': fid2idxs,
'compressed_table': cmp_table,
# sample metadata
'sample_metadata_columns': sm_cols,
'compressed_sample_metadata': compressed_sm,
# feature metadata
'feature_metadata_columns': fm_cols,
'split_taxonomy_columns': self.tax_cols,
'compressed_tip_metadata': compressed_tm,
'compressed_int_metadata': compressed_im,
# Emperor integration
'emperor_div': '',
'emperor_require_logic': '',
'emperor_style': '',
'emperor_base_dependencies': '',
'emperor_classes': ''
}
if self._emperor is not None:
data_to_render.update(self._scavenge_emperor())
return data_to_render
