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

# In[1]:


get_ipython().run_line_magic('load_ext', 'autoreload')
get_ipython().run_line_magic('autoreload', '2')

import matplotlib.pyplot as plt
import seaborn as sns
import numpy as np
import pandas as pd

sns.set(style='ticks', context='paper', rc={'font.sans-serif':'Arial', 'pdf.fonttype': 42})

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

import flotilla


flotilla_dir = '/projects/ps-yeolab/obotvinnik/flotilla_projects'

study = flotilla.embark('singlecell_pnm_figure1_supplementary_post_splicing_filtering', flotilla_dir=flotilla_dir)

not_outliers = study.splicing.singles.index.difference(study.splicing.outliers.index)
folder = 'pdfs' #'/home/obotvinnik/Dropbox/figures2/singlecell_pnm/figure2_modalities/bayesian'
get_ipython().system(' mkdir $folder')


# In[2]:


geo_folder = '/home/obotvinnik/projects/singlecell_pnms/data/geneyeo'


# In[3]:


cd $geo_folder


# In[4]:


import os
import glob

r1s = pd.Series(glob.glob('*R1.fastq.gz'), name='R1').to_frame()
r1s.head()


# In[5]:


r1s['sample_id'] = r1s['R1'].map(lambda x: '_'.join(x.split('_')[:2]))
r1s.head()


# In[6]:


study.expression.data.shape


# In[7]:


r1s_valid = r1s.loc[r1s['sample_id'].isin(study.expression.data.index)]
print(r1s_valid.shape)
r1s_valid.head()


# In[8]:


r1s_invalid = r1s.loc[~r1s['sample_id'].isin(study.expression.data.index)]
r1s_invalid.shape


# In[9]:


from __future__ import print_function


# In[10]:


ls


# In[11]:


for r1 in r1s_invalid['R1']:
    print(r1)
    r2 = r1.replace('R1', 'R2')
    get_ipython().system(' rm -rf $r1')
    get_ipython().system(' rm -rf $r2')


# In[12]:


pwd


# In[13]:


# cd ..


# In[14]:


# cd olga.botvinnik@gmail.com/


# In[15]:


all_fastqs = pd.Series(glob.glob('*.fastq.gz'), name='fastq').to_frame()
all_fastqs = all_fastqs.sort_values('fastq')
all_fastqs.head()


# In[16]:


md5 = get_ipython().getoutput(' md5sum CVN_01_R1.fastq.gz')


# In[17]:


md5[0].split()


# In[18]:


all_fastqs['md5'] = None


# In[19]:


r2s = pd.Series(glob.glob('*R2.fastq.gz')).to_frame()
r2s.head()


# In[20]:


for i, row in all_fastqs.iterrows():
    filename = row['fastq']
    md5 = get_ipython().getoutput(' md5sum $filename')
    md5 = md5[0].split()[0]
    all_fastqs.loc[i, 'md5'] = md5
all_fastqs.head()


# In[ ]:


all_fastqs.to_csv('fastq_md5.csv', index=False)


# In[ ]:


all_fastqs.to_clipboard()


# ## Get checksums of processed data

# In[2]:


cd /projects/ps-yeolab/obotvinnik/flotilla_projects/singlecell_pnm_figure4_voyages/


# In[3]:


get_ipython().system(' md5sum *csv')


# ## Make soft links to everything

# In[5]:


cd ~/projects/singlecell_pnms/data/geneyeo/


# In[6]:


get_ipython().system(' ln -s /projects/ps-yeolab/obotvinnik/flotilla_projects/singlecell_pnm_figure4_voyages/*csv .')


# In[8]:


get_ipython().system(' ln -s /projects/ps-yeolab/genomes/hg19/hg19_phastcons_placental_mammal.wig .')


# In[9]:


get_ipython().system(' md5sum hg19_phastcons_placental_mammal.wig')


# In[10]:


get_ipython().system(' ln -s /projects/ps-yeolab/obotvinnik/singlecell_pnms/csvs_for_paper/splicing_feature_data/alternative/exons.bed alternative_exons.bed')


# In[11]:


get_ipython().system(' ln -s /projects/ps-yeolab/obotvinnik/singlecell_pnms/csvs_for_paper/splicing_feature_data/constitutive/exons.bed constitutive_exons.bed')


# In[12]:


get_ipython().system(' md5sum *.bed')


# In[13]:


get_ipython().system(' ls -1 | wc -l')


# In[ ]: