handle empty plotvals (in yet another place...)

and fiddle with sklearn/numpy imports to avoid warning

python/mds.py

@@ -6,6 +6,7 @@
 import numpy
 import itertools
 import subprocess
+import warnings
 
 import utils
 
@@ -233,8 +234,10 @@ def run_bios2mds(n_components, n_clusters, seqfos, base_workdir, seed, aligned=F
 # ----------------------------------------------------------------------------------------
 def run_sklearn_mds(n_components, n_clusters, seqfos, seed, reco_info=None, region=None, aligned=False, n_init=4, max_iter=300, eps=1e-3, n_jobs=-1, plotdir=None):
     print '%s not testing this after moving these imports down here' % utils.color('red', 'hey')
-    from sklearn import manifold  # these are both slow af to import, even on local ssd
-    from sklearn.cluster import KMeans
+    with warnings.catch_warnings():  # NOTE not sure this is actually catching the warnings
+        warnings.simplefilter('ignore')  # numpy is complaining about how sklearn is importing something, and I really don't want to *@*($$ing hear about it
+        from sklearn import manifold  # these are both slow af to import, even on local ssd
+        from sklearn.cluster import KMeans
 
     if len(set(sfo['name'] for sfo in seqfos)) != len(seqfos):
         raise Exception('duplicate sequence ids in <seqfos>')

python/plotting.py

@@ -1153,6 +1153,8 @@ def get_xval_dict(uids, xkey):
     # ----------------------------------------------------------------------------------------
     def getbounds(xkey):
         all_xvals = [x for c in sorted_clusters for x in get_xval_list(c, xkey)]
+        if len(all_xvals) == 0:
+            return None
         bounds = [f(all_xvals) for f in [min, max]]
         if global_max_vals is not None and xkey in global_max_vals:
             bounds[1] = global_max_vals[xkey]
@@ -1243,6 +1245,8 @@ def add_hist(xkey, xvals, yval, iclust, cluster, median_x1, fixed_x1max, base_al
     xbounds = {x1key : getbounds(x1key)}  # these are the smallest/largest x values in any of <sorted_clusters>, whereas <high_x_val> is a fixed calling-fcn-specified value that may be more or less (kind of wasteful to get all the x vals here and then also in the main loop)
     if x2key is not None:
         xbounds[x2key] = getbounds(x2key)
+    if any(xbounds[xk] is None for xk in xbounds):
+        return 'no values' if high_x_val is None else high_x_clusters  # 'no values' isn't really a file name, it just shows up as a dead link in the html
     fixed_xmax = high_x_val if high_x_val is not None else xbounds[x1key][1]  # xmax to use for the plotting (ok now there's three max x values, this is getting confusing)
 
     if debug:

python/treeutils.py

@@ -16,10 +16,9 @@
 import dendropy
 import time
 import math
-from sklearn import tree as sktree
-from sklearn import ensemble
 import yaml
 import pickle
+import warnings
 if StrictVersion(dendropy.__version__) < StrictVersion('4.0.0'):  # not sure on the exact version I need, but 3.12.0 is missing lots of vital tree fcns
     raise RuntimeError("dendropy version 4.0.0 or later is required (found version %s)." % dendropy.__version__)
 
@@ -87,6 +86,11 @@ def dtrfname(dpath, cg):
 
 # ----------------------------------------------------------------------------------------
 def train_dtr(cgroup, dtrfo, dmodels, outdir, min_samples_leaf=5, max_depth=10, n_estimators=10, dump_training_data=False):
+    with warnings.catch_warnings():  # NOTE not sure this is actually catching the warnings
+        warnings.simplefilter('ignore', category=DeprecationWarning)  # numpy is complaining about how sklearn is importing something, and I really don't want to *@*($$ing hear about it
+        from sklearn import tree as sktree
+        from sklearn import ensemble
+
     print '    %s' % cgroup.replace('-', ' ')
 
     base_regr = sktree.DecisionTreeRegressor(min_samples_leaf=min_samples_leaf, max_depth=max_depth)