Merge pull request #48 from yuukiiwa/gtf

gtf option, fastdataprep, and dataprep intermediate file removal

setup.py

@@ -26,7 +26,8 @@
             'scipy>=1.4.1',
             'PyYAML',
             'h5py>=2.10.0',
-            'pyensembl>=1.8.5'
+            'pyensembl>=1.8.5',
+            'ujson>=4.0.1'
             ],
     python_requires=">=3.8",
     entry_points={'console_scripts': ["xpore-dataprep={}.scripts.dataprep:main".format(__pkg_name__),

xpore/diffmod/configurator.py

@@ -51,7 +51,7 @@ def get_method(self):
 
         method.setdefault('name', 'gmm')
         method.setdefault('max_iters', 500)
-        method.setdefault('stopping_criteria', 0.0001)
+        method.setdefault('stopping_criteria', 0.00001)
         method.setdefault('compute_elbo', True)
         method.setdefault('verbose', False)
         method.setdefault('update', ['z','y','w','mu_tau'])

xpore/scripts/diffmod.py

@@ -3,7 +3,7 @@
 import pandas
 import os
 import multiprocessing 
-import json
+import ujson
 from collections import defaultdict
 import csv
 
@@ -187,7 +187,7 @@ def main():
                     json_str = f_data[run_name].read(pos_end-pos_start)
                     # print(json_str[:50])
                     # json_str = '{%s}' %json_str # used for old dataprep
-                    data_dict[(condition_name,run_name)] = json.loads(json_str) # A data dict for each gene.
+                    data_dict[(condition_name,run_name)] = ujson.loads(json_str) # A data dict for each gene.
 
         # tmp
         out_paths['model_filepath'] = os.path.join(paths['models'],'%s.hdf5' %idx)

xpore/scripts/helper.py

@@ -31,6 +31,9 @@ def __exit__(self, exc_type, exc_value, exc_traceback):
     def close(self):
         self._handle.close()
 
+    def readline(self):
+        self._handle.readline()
+
     def __iter__(self):
         return self
 

xpore/utils/stats.py

@@ -18,7 +18,13 @@ def z_test(y1, y2, n1, n2):
 def calc_prob_overlapping(means, variances):
     sigma1, sigma2 = np.sqrt(variances)
     mu1, mu2 = means
-    return NormalDist(mu=mu1, sigma=sigma1).overlap(NormalDist(mu=mu2, sigma=sigma2))
+    ### the bug causing entry had sigma1 == sigma2 and mu1 == mu2
+    #print("sigma1: ",sigma1,"sigma2: ",sigma2)
+    #print("mu1: ",mu1,"mu2: ",mu2)
+    if sigma1 != sigma2 and mu1 != mu2:
+        return NormalDist(mu=mu1, sigma=sigma1).overlap(NormalDist(mu=mu2, sigma=sigma2))
+    else:
+        return np.nan,[np.nan,np.nan,np.nan,np.nan]
 
 class NormalDist(object):
     "Modified from the library statistics.py (Authors)(https://github.com/python/cpython/blob/3.8/Lib/statistics.py) which is under GPL-compatible license."