Update for Python 3

README.md

@@ -12,7 +12,7 @@ Code repository corresponding to paper "Detecting differential growth of microbi
 * python (virtualenv recommended)
 
 #### Download this repository
-`git clone https://github.com/ptonner/gp_growth_phenotype.git`
+`git clone https://github.com/SirRujak/gp_growth_phenotype.git`
 
 #### Setup a local python environment (optional)
 `virtualenv .`
@@ -24,6 +24,8 @@ Code repository corresponding to paper "Detecting differential growth of microbi
 
 ## Examples
 
+For a basic example that does not use jupyter, look at example.py in the notebooks folder.
+
 Input to the B-GREAT method should come from two dataframes, *data* (n x p) and *meta* (p x k) where n, p, and k are:
 * n: number of time-points
 * p: number of time-course measurements (replicates)

bgreat.py

@@ -1,8 +1,12 @@
+from __future__ import absolute_import
+from __future__ import print_function
 import matplotlib.pyplot as plt
 import numpy as np
 import scipy, GPy, os, ast
 import pandas as pd
 import matplotlib as mpl
+from six.moves import range
+from six.moves import zip
 
 meta = data = parent = condition = control = None
 
@@ -14,13 +18,13 @@
 
 def plotSamples(samples,x=None,color='b',colors=None,plotMethod=None,label=None,*args,**kwargs):
 	if x is None:
-		x = range(samples.shape[0])
+		x = list(range(samples.shape[0]))
 	if colors is None:
 		colors = [color]*samples.shape[1]
 	if plotMethod is None:
 		plotMethod = plt.plot
 
-	for i,c in zip(range(samples.shape[1]),colors):
+	for i,c in zip(list(range(samples.shape[1])),colors):
 		if not label is None:
 			plotMethod(x,samples[:,i],color=c,label=label,*args,**kwargs)
 			label = None
@@ -38,7 +42,7 @@ def setGlobals(_data=None,_meta=None,_parent=None,_condition=None,_control=None)
 		global data
 		data = _data
 
-		data.columns = range(data.shape[1])
+		data.columns = list(range(data.shape[1]))
 
 	if not _meta is None:
 		global meta
@@ -234,7 +238,7 @@ def testMutants(mutants,numPerm=10,timeThin=4,dims=[],nullDim='strain-regression
 	for i,m in enumerate(mutants):
 
 		if (i + 1)%10 == 0:
-			print 1.*i/len(mutants),m
+			print(1.*i/len(mutants),m)
 
 		select = ((meta.Condition==control) | (meta.Condition==condition)) & (meta.strain.isin([parent,m]))
 		if m in results:

gp_growth/analysis.py

@@ -1,9 +1,13 @@
+from __future__ import absolute_import
+from __future__ import print_function
 import pandas as pd
 import numpy as np
-from ConfigParser import ConfigParser
+from six.moves.configparser import ConfigParser
 from registry import registry
 from gp_growth import storage
 import os, gc
+from six.moves import range
+from six.moves import zip
 
 class Analysis():
 
@@ -111,16 +115,16 @@ def run(self,verbosity=None):
 					continue	
 
 				# build a dictionary of key-value pairs for selecting data
-				kwargs = dict(zip(self.groupby,vals))
+				kwargs = dict(list(zip(self.groupby,vals)))
 
 				temp = self.data.select(**kwargs)
 
 				if verbosity > 0:
-					print kwargs, temp.key.shape[0]
+					print(kwargs, temp.key.shape[0])
 
 				if temp.key.shape[0] > self.maxSize:
 					if verbosity > 0:
-						print "Number of samples to large (%d > %d)" % (temp.key.shape[0],self.maxSize)
+						print("Number of samples to large (%d > %d)" % (temp.key.shape[0],self.maxSize))
 					continue
 
 				regression = temp.getData("gp")
@@ -133,7 +137,7 @@ def run(self,verbosity=None):
 				# del temp
 				collected = gc.collect()
 				if verbosity > 0:
-					print "Garbage collector: collected %d objects." % (collected)
+					print("Garbage collector: collected %d objects." % (collected))
 
 if __name__ == "__main__":
 	import sys, getopt

gp_growth/categorical.py

@@ -1,3 +1,5 @@
+from __future__ import absolute_import
+from __future__ import print_function
 from GPy.kern import Kern,RBF
 from GPy.core.parameterization.transformations import Logexp
 from GPy.core.parameterization import Param
@@ -37,15 +39,15 @@ def update_gradients_full(self, dL_dK, X, X2=None):
 			self.variance.gradient = np.nansum(dL_dK * np.where(X == X2.T,1,0))
 
 	def update_gradients_diag(self, dL_dKdiag, X):
-		print "update_gradients_diag"
+		print("update_gradients_diag")
 		self.variance.gradient = np.sum(dL_dKdiag * np.ones(X.shape[0]))
 
 	def gradients_X(self, dL_dK, X, X2):
-		print "gradients_X"
+		print("gradients_X")
 		if X2 is None:
 			X2 = X
 		return dL_dK * np.where(X == X2.T,1,0)
 
 	def gradients_X_diag(self, dL_dKdiag, X):
-		print "gradients_X_diag"
+		print("gradients_X_diag")
 		return dL_dKdiag * np.ones(X.shape[0])

gp_growth/changepoint.py

@@ -1,9 +1,12 @@
+from __future__ import absolute_import
+from __future__ import print_function
 from gp_growth import factory
 from gp_growth.categorical import Categorical
 import GPy
 from GPy.core.parameterization import Param
 from GPy.core.parameterization.transformations import Logexp
 import numpy as np
+from six.moves import range
 
 CP_TOL = 1e-5
 
@@ -130,7 +133,7 @@ class ChangepointCross(GPy.kern._src.kern.CombinationKernel):
     """Kernel for points across a changepoint. K(X,Y) = kc * K1(X,cp) * K2(Y,cp) """
 
     def __init__(self,cp,kc = 1,k=1,**kwargs):
-        ad = [0] + range(2,k+1)
+        ad = [0] + list(range(2,k+1))
 
         super(ChangepointCross,self).__init__([GPy.kern.RBF(k,active_dims=ad,ARD=True,**kwargs),GPy.kern.RBF(k,active_dims=ad,ARD=True,**kwargs)],"changepoint")
         self.k = k
@@ -157,7 +160,7 @@ def Kdiag(self,X):
         return self.kc * self.parts[0].Kdiag(X) * self.parts[1].Kdiag(X)
 
     def update_gradients_full(self, dL_dK, X, X2=None):
-        print "cp_cross update_gradients_full"
+        print("cp_cross update_gradients_full")
         k = self.K(X,X2)*dL_dK
 #         try:
         for p in self.parts:
@@ -208,7 +211,7 @@ def buildKernelPreprocess(self,):
     def buildKernel(self,):
 
         k = len(self.inputDimensions) - 1
-        ad = [0] + range(2,k+1)
+        ad = [0] + list(range(2,k+1))
 
         cp = self.cp
         if self.normalize:

gp_growth/changepoint_multidim.py

@@ -1,10 +1,13 @@
+from __future__ import absolute_import
+from __future__ import print_function
 from gp_growth import factory
 from gp_growth.categorical import Categorical
 import GPy
 from GPy.core.parameterization import Param
 from GPy.core.parameterization.transformations import Logexp
 from GPy.kern import Kern, RBF
 import numpy as np
+from six.moves import range
 
 CP_TOL = 1e-5
 
@@ -51,7 +54,7 @@ def __init__(self,input_dim=1,k1=None,k2=None,kc=None,xc=None,cpDim=None,*args,*
         self.cpDim = cpDim
         if self.cpDim is None:
             self.cpDim = 0
-        self.otherDims = range(self.input_dim)
+        self.otherDims = list(range(self.input_dim))
         self.otherDims.remove(self.cpDim)
         self.otherDims = np.array(self.otherDims)
 
@@ -298,7 +301,7 @@ class ChangepointCross(GPy.kern._src.kern.CombinationKernel):
     """Kernel for points across a changepoint. K(X,Y) = kc * K1(X,cp) * K2(Y,cp) """
 
     def __init__(self,cp,kc = 1,k=1,**kwargs):
-        ad = [0] + range(2,k+1)
+        ad = [0] + list(range(2,k+1))
 
         super(ChangepointCross,self).__init__([GPy.kern.RBF(k,active_dims=ad,ARD=True,**kwargs),GPy.kern.RBF(k,active_dims=ad,ARD=True,**kwargs)],"changepoint")
         self.k = k
@@ -325,7 +328,7 @@ def Kdiag(self,X):
         return self.kc * self.parts[0].Kdiag(X) * self.parts[1].Kdiag(X)
 
     def update_gradients_full(self, dL_dK, X, X2=None):
-        print "cp_cross update_gradients_full"
+        print("cp_cross update_gradients_full")
         k = self.K(X,X2)*dL_dK
 #         try:
         for p in self.parts:
@@ -376,7 +379,7 @@ def buildKernelPreprocess(self,):
     def buildKernel(self,):
 
         k = len(self.inputDimensions) - 1
-        ad = [0] + range(2,k+1)
+        ad = [0] + list(range(2,k+1))
 
         cp = self.cp
         if self.normalize:

gp_growth/classical.py

@@ -1,3 +1,4 @@
+from __future__ import absolute_import
 import numpy as np
 from scipy.optimize import curve_fit
 
@@ -26,7 +27,7 @@ def meanSquareError(time,od,model=gompertz):
 
 		predict = np.array([gompertz(t,*popt) for t in time])
 		mse = 1./n * sum(((od - predict)**2),)
-	except RuntimeError, e:
+	except RuntimeError as e:
 		mse = np.nan# 1./n * sum(((od - np.mean(od))**2),)
 
 	return mse
@@ -62,7 +63,7 @@ def optimize(time,od,model=gompertz,p0=None,cv=None):
 
 		predict = np.array([model(t,*popt) for t in test_time])
 		mse = 1./test_od.shape[0] * sum(((test_od - predict)**2),)
-	except RuntimeError, e:
+	except RuntimeError as e:
 		m,A,l = np.nan, np.nan, np.nan
 		mse = np.nan
 

gp_growth/data/filter.py

@@ -1,4 +1,5 @@
-import growth
+from __future__ import absolute_import
+from . import growth
 
 class DataFilter(object):
 

gp_growth/data/growth.py

@@ -1,9 +1,14 @@
+from __future__ import absolute_import
+from __future__ import print_function
 import matplotlib.pyplot as plt
 import seaborn as sns
 import os, time, datetime
 import pandas as pd
 import numpy as np
 import itertools
+import six
+from six.moves import range
+from six.moves import zip
 
 class GrowthData(object):
 
@@ -64,7 +69,7 @@ def plot(self,output="",fixed_y=True,title_index=None,colorby=None,fixed_colors=
 			condition = condition.tolist()
 
 		if title_index is None:
-			title_index = range(len(condition))
+			title_index = list(range(len(condition)))
 
 		if newFig is None:
 			newFig = True
@@ -82,9 +87,9 @@ def plot(self,output="",fixed_y=True,title_index=None,colorby=None,fixed_colors=
 		if colorby:
 			categories = tuple(self.key[colorby].unique())
 			if colors is None:
-				color_lookup = dict(zip(categories,sns.color_palette("hls", len(categories))))
+				color_lookup = dict(list(zip(categories,sns.color_palette("hls", len(categories)))))
 			else:
-				color_lookup = dict(zip(categories,colors))
+				color_lookup = dict(list(zip(categories,colors)))
 
 		ncols = min(ncols,groups.ngroups)
 		nrows = 1
@@ -97,12 +102,12 @@ def plot(self,output="",fixed_y=True,title_index=None,colorby=None,fixed_colors=
 			plt.figure(figsize=(ncols*figWidth,nrows*figHeight))
 
 		if groups.ngroups > maxgroups:
-			print "Error, number of groups (%d) is greater than max (%d)." % (groups.ngroups,maxgroups)
+			print("Error, number of groups (%d) is greater than max (%d)." % (groups.ngroups,maxgroups))
 			return
 
-		groupList = list(groups.groups.iteritems())
+		groupList = list(six.iteritems(groups.groups))
 		if groupOrder is None:
-			groupOrder = range(len(groupList))
+			groupOrder = list(range(len(groupList)))
 		for i,j in enumerate(groupOrder):
 			val = groupList[j]
 
@@ -119,10 +124,10 @@ def plot(self,output="",fixed_y=True,title_index=None,colorby=None,fixed_colors=
 
 			if not fixed_colors:
 				categories = tuple(temp_key[colorby].unique())
-				color_lookup = dict(zip(categories,sns.color_palette("hls", len(categories))))
+				color_lookup = dict(list(zip(categories,sns.color_palette("hls", len(categories)))))
 
 			if colorby:
-				color_usage = dict(zip(categories,[False]*len(categories)))
+				color_usage = dict(list(zip(categories,[False]*len(categories))))
 
 			#print temp_key.Well,temp_data.head()
 
@@ -132,7 +137,7 @@ def plot(self,output="",fixed_y=True,title_index=None,colorby=None,fixed_colors=
 				ylim = (min(temp_data.min()),max(temp_data.max()))
 
 			if colorby:
-				for well,temp_od in temp_data.iteritems():
+				for well,temp_od in six.iteritems(temp_data):
 
 					#remove na's
 					select = ~temp_od.isnull()
@@ -152,7 +157,7 @@ def plot(self,output="",fixed_y=True,title_index=None,colorby=None,fixed_colors=
 				if len(categories) <= max_legend:
 					plt.legend(loc="best",**legend_kwargs)
 			else:
-				for well,temp_od in temp_data.iteritems():
+				for well,temp_od in six.iteritems(temp_data):
 
 					#remove na's
 					select = ~temp_od.isnull()
@@ -263,7 +268,7 @@ def _stack_data(self,meta=None,groupby=None,thinning=None):
 			x = []
 			y = []
 
-			for vals,ind in g.groups.iteritems():
+			for vals,ind in six.iteritems(g.groups):
 				select = reg[groupby] == vals
 				temp = reg[select]
 				x.append(reg[['time']+meta])
@@ -308,7 +313,7 @@ def select(self,**kwargs):
 
 		selection = [True]*self.key.shape[0]
 
-		for k,v in kwargs.iteritems():
+		for k,v in six.iteritems(kwargs):
 			if k in key_copy.columns:
 				# if v is np.nan:
 				# if np.isnan(v):
@@ -317,7 +322,7 @@ def select(self,**kwargs):
 					if np.isnan(v):
 						selection = np.all((selection,key_copy[k].isnull()),0)
 						checked = True
-				except TypeError,e:
+				except TypeError as e:
 					pass
 				if not checked:
 					selection = np.all((selection,key_copy[k]==v),0)
@@ -368,7 +373,7 @@ def poly_scale(self,p,ind=None,groupby=None):
 
 		time = self.data.time.iloc[:ind]
 
-		for k,index in group.groups.iteritems():
+		for k,index in six.iteritems(group.groups):
 		    temp = self.data.loc[:,index]
 		    od = temp.values[:ind,:].ravel()
 
@@ -380,13 +385,13 @@ def poly_scale(self,p,ind=None,groupby=None):
 def _parse_time(t):
 	try:
 		return time.struct_time(time.strptime(t,'%H:%M:%S'))
-	except ValueError, e:
+	except ValueError as e:
 		try:
 			t = time.strptime(t,'%d %H:%M:%S')
 			t = list(t)
 			t[2]+=1
 			return time.struct_time(t)
-		except ValueError, e:
+		except ValueError as e:
 			raise Exception("Time format unknown")
 
 def _expand_data_row(r,thinning=0):