#!/usr/bin/env python
# coding: utf-8

# # Introduction

# We provide a set of notebooks to show how the GDSCTools package can be used in ipython / ipython notebook.
# 
# The source code is available on github https://github.com/CancerRxGene/gdsctools
# Would you have any issues (bug related), please fill an issue here https://github.com/CancerRxGene/gdsctools/issues
# 
# In this notebook, we will simply give a flavour of what can be done. Other notebooks will provide more detailed examples. 
# 
# 
# Documentation is also available for users and developers in a dedicated entry page on Pypi and http://gdsctools.readthedocs.org
# 
# <hr>
# 
# <div id="content">
# <p>**Other notebooks:**</p><ul id="toc"/>
# <ul>
# <li><a href="2. Data.ipynb">About the input data</a></li>
# <li><a href="3. One Drug Analysis.ipynb">Analysis one drug only and volcano plot</a></li>
# <li><a href="4. One Drug One Feature Analysis.ipynb">Analysis of one drug one feature (Association) and
# boxplots</a></li>
# <li><a href="5. Report.ipynb">Create HTML report</li>
# <li><a href="6. Settings.ipynb">Settings of the analysis</li>
# <li><a href="7. Cancer Specific Analysis.ipynb">Cancer specific analysis</li>
# </ul>
# </div>
# 
# 

# ### Overview

# The goal of this package is to provide tools related to the GDSC project 
# (Genomics of Drug Sensitivity in Cancer) http://www.cancerrxgene.org/

# Currently, GSDSTools provides functionalities to identify associations between drugs and genomic features across a set of cell lines 

# The genomic features are provided within the packages. Users need to provide IC50 for a set of drugs and a set of cell lines

# We provide an example to play with. First let us get this IC50 test file and ad it

# In[1]:


get_ipython().run_line_magic('pylab', 'inline')

from gdsctools import ic50_test, GenomicFeatures, genomic_features_test


# In[2]:


print(ic50_test)


# This is just a file with a location and description. It can be read using
# the IC50 class

# In[3]:


from gdsctools import IC50
data = IC50(ic50_test)
print(data)


# As you can see, it contains 11 drugs across 988 cell lines

# Similarly, there is a genomic feature data set provided, which can be read 
# with the  GenomicFeatures class

# In[4]:


gf = GenomicFeatures(genomic_features_test)
print(gf)


# This file is going to be downloaded automatically when an analysis 
# is performed. However, you may provide your own file.

# Let us now perform the analysis using the ANOVA class

# In[5]:


from gdsctools import ANOVA


# In[6]:


an = ANOVA(data, genomic_features=genomic_features_test.filename)


# In[7]:


print(an)


# so, we have 11 drugs, 677 features across 988 cell lines (27 tissues). This 
# is a PANCAN analysis (across several cancer cell types).
# 
# We can analysis the entire data set, which takes some time (still reasonable; about 1 minute dependiing on your system). 

# In[8]:


results = an.anova_all()


# All results are now in the new variable <b>results</b>, which can be looked at. This is a dataframe formatted variable using Pandas library. Each association can be accessed to using a unique identifier from 0 to the length of the dataframe:

# In[9]:


results.df.loc[0]


# As an example, we can plot the histogram of the FDR columns:

# In[10]:


get_ipython().run_line_magic('pylab', 'inline')
results.df['ANOVA_FEATURE_FDR'].hist(bins=20)


# In the next notebooks, we will now investigate more precisely 
# - the input data sets
# - the analysis and in particular how to look at
#     - one association
#     - associations for a given drug
#     - all associations (what we did here when we called anova_all() function)
# - How to generate HTML reports
# - The settings
#     

# In[ ]: