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

# ## GenomeTools Scripts
# 
# ```
# Author: Florian Wagner
# Email: florian.wagner@duke.edu
# ```
# This notebook demonstrates the use of the scripts `extact_protein_coding_genes.py` and `extract_entrez2gene.py` from the GenomeTools package.

# In[1]:


# get genometools version
from pkg_resources import require

print 'Package versions'
print '----------------'
print require('genometools')[0]


# ### Running `extract_protein_coding_genes.py`

# In[2]:


gene_annotation_file = 'Homo_sapiens.GRCh38.79.gtf.gz'
protein_coding_gene_file = 'protein_coding_genes_human.tsv.gz'


# In[3]:


get_ipython().system('curl -o "$gene_annotation_file"          "ftp://ftp.ensembl.org/pub/release-79/gtf/homo_sapiens/Homo_sapiens.GRCh38.79.gtf.gz"')


# In[4]:


# reading annotation file from stdin
get_ipython().system('gunzip -c "$gene_annotation_file" | extract_protein_coding_genes.py -q -a - -o - | gzip > "$protein_coding_gene_file"')
get_ipython().system('gunzip -c "$protein_coding_gene_file" | head -n 10')


# In[5]:


# alternatively: reading the annotation file directly
get_ipython().system('extract_protein_coding_genes.py -a "$gene_annotation_file" -o - | gzip > "$protein_coding_gene_file"')


# ### Running `extract_gene2entrez.py`

# In[6]:


gene2accession_file = 'gene2accession_2015-05-26_human.tsv.gz'
entrez2gene_file = 'entrez2gene_human.tsv'


# In[7]:


get_ipython().system('curl -L -o "$gene2accession_file"          "https://www.dropbox.com/s/ggjrvnigtrfue3x/gene2accession_human_2015-05-26.tsv.gz?dl=1"')


# In[8]:


get_ipython().system('extract_entrez2gene.py -f "$gene2accession_file" -o "$entrez2gene_file"')
get_ipython().system('head -n 10 "$entrez2gene_file"')