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

# # gathergrams - display gather results

# In[1]:


get_ipython().run_line_magic('matplotlib', 'inline')
import pylab
import numpy
import pandas as pd


# In[2]:


sample_id = 'SRR606249'
df = pd.read_csv(f'outputs/{sample_id}.gather.csv')

def fix_name(x):
    return "_".join(x.split('_')[:2]).split('.')[0]

df['name'] = df['name'].apply(fix_name)
df[:4]


# # Figure 1: gathergram

# In[3]:


pylab.figure(num=None, figsize=(8, 6))

pylab.plot(df.index, df.intersect_bp / 1e6, 'x', label='all hashes classified to this genome')
pylab.plot(df.index, df.unique_intersect_bp / 1e6, 'o', label='hashes specific to this genome')
#pylab.plot(df.index, df.remaining_bp / 1e6, '-', label='unclassified bp remaining')

pylab.xlabel('genome gather rank')
pylab.ylabel('num hashes (millions)')
pylab.legend(loc='upper right')
pylab.title('hu-s1 gather x genbank')

pylab.savefig(f'/tmp/gathergram-{sample_id}.pdf')


# In[ ]: