Merge pull request #30 from BojarLab/dev

merge into master for v1.0.1

build/lib/glycowork/glycan_data/loader.py

@@ -1,3 +1,4 @@
+import numpy as np
 import pandas as pd
 import re
 import os
@@ -188,6 +189,52 @@ def multireplace(string, remove_dic):
   return string
 
 
+def fast_two_sum(a, b):
+  """Assume abs(a) >= abs(b)"""
+  x = int(a) + int(b)
+  y = b - (x - int(a))
+  return [x] if y == 0 else [x, y]
+
+
+def two_sum(a, b):
+  """For unknown order of a and b"""
+  x = int(a) + int(b)
+  y = (a - (x - int(b))) + (b - (x - int(a)))
+  return [x] if y == 0 else [x, y]
+
+
+def expansion_sum(*args):
+  """For the expansion sum of floating points"""
+  g = sorted(args, reverse = True)
+  q, *h = fast_two_sum(np.array(g[0]), np.array(g[1]))
+  for val in g[2:]:
+    z = two_sum(q, np.array(val))
+    q, *extra = z
+    if extra:
+      h += extra
+  return [h, q] if h else q
+
+
+def hlm(z):
+  """Hodges-Lehmann estimator of the median"""
+  z = np.array(z)
+  zz = np.add.outer(z, z)
+  zz = zz[np.tril_indices(len(z))]
+  return np.median(zz) / 2
+
+
+def update_cf_for_m_n(m, n, MM, cf):
+  """Constructs cumulative frequency table for experimental parameters defined in the function 'jtkinit'"""
+  P = min(m + n, MM)
+  for t in range(n + 1, P + 1):  # Zero-based offset t
+    for u in range(MM, t - 1, -1):  # One-based descending index u
+      cf[u] = expansion_sum(cf[u], -cf[u - t])  # Shewchuk algorithm
+  Q = min(m, MM)
+  for s in range(1, Q + 1): # Zero-based offset s
+    for u in range(s, MM + 1):  # One-based descending index u
+      cf[u] = expansion_sum(cf[u], cf[u - s])  # Shewchuk algorithm
+
+
 def build_custom_df(df, kind = 'df_species'):
   """creates custom df from df_glycan\n
   | Arguments:

build/lib/glycowork/motif/analysis.py

@@ -16,7 +16,7 @@
 from sklearn.decomposition import PCA
 
 from glycowork.glycan_data.loader import lib, df_species, unwrap, motif_list
-from glycowork.motif.processing import cohen_d, mahalanobis_distance, mahalanobis_variance, variance_stabilization, impute_and_normalize, variance_based_filtering
+from glycowork.motif.processing import cohen_d, mahalanobis_distance, mahalanobis_variance, variance_stabilization, impute_and_normalize, variance_based_filtering, jtkdist, jtkinit, MissForest, jtkx
 from glycowork.motif.annotate import annotate_dataset, quantify_motifs, link_find, create_correlation_network
 from glycowork.motif.graph import subgraph_isomorphism
 
@@ -849,3 +849,36 @@ def get_time_series(df, impute = True, motifs = False, feature_set = ['known', '
     res = pd.DataFrame(res, columns = ['Glycan', 'Change', 'p-val'])
     res['corr p-val'] = multipletests(res['p-val'], method = 'fdr_bh')[1]
     return res.sort_values(by = 'corr p-val')
+
+
+def get_jtk(df, timepoints, replicates, periods, interval, motifs = False, feature_set = ['known', 'exhaustive', 'terminal']):
+    """Wrapper function running the analysis \n
+    | Arguments:
+    | :-
+    | df (pd.DataFrame): A dataframe containing data for analysis.
+    |   (column 0 = molecule IDs, then arranged in groups and by ascending timepoints)
+    | timepoints (int): number of timepoints in the experiment.
+    | replicates (int): number of replicates per timepoints.
+    | periods (int): number of timepoints per cycle.
+    | interval (int): units of time (Arbitrary units) between experimental timepoints.
+    | motifs (bool): a flag for running structural of motif-based analysis (True = run motif analysis); default:False.
+    | feature_set (list): which feature set to use for annotations, add more to list to expand; default is ['exhaustive','known']; options are: 'known' (hand-crafted glycan features), 'graph' (structural graph features of glycans), 'exhaustive' (all mono- and disaccharide features), 'terminal' (non-reducing end motifs), and 'chemical' (molecular properties of glycan)\n
+    | Returns:
+    | :-
+    | Returns a pandas dataframe containing the adjusted p-values, and most important waveform parameters for each
+    | molecule in the analysis.
+    """
+    param_dic = {"GRP_SIZE": [], "NUM_GRPS": [], "MAX": [], "DIMS": [], "EXACT": bool(True),
+                 "VAR": [], "EXV": [], "SDV": [], "CGOOSV": []}
+    param_dic = jtkdist(timepoints, param_dic, replicates)
+    param_dic = jtkinit(periods, param_dic, interval, replicates)
+    mf = MissForest()
+    df.replace(0, np.nan, inplace = True)
+    df = mf.fit_transform(df)
+    if motifs:
+        df = quantify_motifs(df.iloc[:, 1:], df.iloc[:, 0].values.tolist(), feature_set).T
+    res = df.apply(jtkx, param_dic = param_dic, axis = 1)
+    JTK_BHQ = pd.DataFrame(sm.stats.multipletests(res[0], method = 'fdr_bh')[1])
+    Results = pd.concat([res.iloc[:, 0], JTK_BHQ, res.iloc[:, 1:]], axis = 1)
+    Results.columns = ['Molecule_Name', 'BH_Q_Value', 'Adjusted_P_value', 'Period_Length', 'Lag_Phase', 'Amplitude']
+    return Results

build/lib/glycowork/motif/draw.py

@@ -6,11 +6,9 @@
 import copy
 
 try:
-  import cairosvg
   import drawsvg as draw
 except ImportError:
   raise ImportError("<draw dependencies missing; did you do 'pip install glycowork[draw]'?>")
-import networkx as nx
 import numpy as np
 import sys
 import re
@@ -2147,7 +2145,11 @@ def calculate_degree(y1, y2, x1, x2):
         with open(filepath, 'w') as f:
           f.write(data)
       elif 'pdf' in filepath:
-        cairosvg.svg2pdf(bytestring = data, write_to = filepath)
+        try:
+          from cairosvg import svg2pdf
+          svg2pdf(bytestring = data, write_to = filepath)
+        except:
+          raise ImportError("You're missing some draw dependencies. Either use .svg or head to https://bojarlab.github.io/glycowork/examples.html#glycodraw-code-snippets to learn more.")
   return d2
 
 
@@ -2244,11 +2246,15 @@ def annotate_figure(svg_input, scale_range = (25, 80), compact = False, glycan_s
   svg_tmp += '</svg>'
 
   if filepath:
+    try:
+      from cairosvg import svg2pdf, svg2svg, svg2png
       if filepath.split('.')[-1] == 'pdf':
-        cairosvg.svg2pdf(bytestring = svg_tmp, write_to = filepath, dpi = 300)
+        svg2pdf(bytestring = svg_tmp, write_to = filepath, dpi = 300)
       elif filepath.split('.')[-1] == 'svg':
-        cairosvg.svg2svg(bytestring = svg_tmp, write_to = filepath, dpi = 300)
+        svg2svg(bytestring = svg_tmp, write_to = filepath, dpi = 300)
       elif filepath.split('.')[-1] == 'png':
-        cairosvg.svg2png(bytestring = svg_tmp, write_to = filepath, dpi = 300) 
+        svg2png(bytestring = svg_tmp, write_to = filepath, dpi = 300)
+    except:
+      raise ImportError("You're missing some draw dependencies. Either don't use filepath or head to https://bojarlab.github.io/glycowork/examples.html#glycodraw-code-snippets to learn more.")
   else:
-      return svg_tmp
+    return svg_tmp