In [1]:

    

import pandas as pd
import sys
import os, os.path

sys.path.append('/home/will/PatientPicker/')


    

In [2]:

    

import LoadingTools


    

In [3]:

    

redcap_data = LoadingTools.load_redcap_data().set_index(['Patient ID', 'VisitNum'])
cyto_data = pd.read_csv('/home/will/HIVSystemsBio/NewCytokineAnalysis/CytoRawData.csv', sep='\t')
cyto_data['HasCyto'] = True
has_cyto = cyto_data.groupby(['Patient ID', 'VisitNum'])[['HasCyto']].all()


    

In [19]:

    

tmp = pd.read_csv('/home/will/HIVSystemsBio/NewCytokineAnalysis/CytoPatData.csv', sep='\t')
ages = tmp.groupby('Patient ID')['Age'].first()
print ages.max(), ages.min()


    

    




71 20

In [4]:

    

import glob
files = glob.glob('/home/will/HIVReportGen/Data/PatientFasta/*.fasta')
seqs = []
for f in files:
    fname = f.split('/')[-1]
    try:
        pid, vn, prot = fname.split('.')[0].split('-', 2)
    except ValueError:
        print fname
    seqs.append((pid, vn, prot, 1))
    
df = pd.DataFrame(seqs, columns = ['Patient ID', 'VisitNum', 'Prot', 'HasSeq'])
has_seq = pd.pivot_table(df, rows = ['Patient ID', 'VisitNum'], cols = 'Prot', values='HasSeq')


    

In [5]:

    

redcap_data = pd.merge(redcap_data, has_cyto,
                       left_index=True, right_index=True,
                       how='outer')
redcap_data = pd.merge(redcap_data, has_seq,
                       left_index=True, right_index=True,
                       how='outer')
redcap_data = redcap_data.drop(['VisitNum', 'Patient ID'], axis=1)


    

In [6]:

    

import numpy
def safe_mean(col):
    ncol = col.dropna()
    if len(ncol) == 0:
        return np.nan
    return ncol.mean()

def safe_col_apply_mean(col, func, indf):
    return indf[col].dropna().map(func).mean()


    

In [7]:

    

from functools import partial
seq_cols = list(has_seq.columns)
test_drug_cols = [col for col in redcap_data.columns if col.startswith('Test-')]
admit_drug_cols = [col for col in redcap_data.columns if col.startswith('Admit-')]
race_cols = [col for col in redcap_data.columns if col.startswith('Race-')]

mean_cols = test_drug_cols+admit_drug_cols+seq_cols+race_cols
agg_dict = dict([(col, safe_mean) for col in mean_cols])
agg_dict['VisitNum'] = 'count'

cut_list = [('%LowVL', 'Latest viral load', 50),
              ('%LowCD4', 'Latest CD4 count (cells/uL)', 200)]


    

In [8]:

    

pat_sum = redcap_data.reset_index().groupby('Patient ID').agg(agg_dict)
for ncol, tcol, cut in cut_list:
    pat_sum[ncol] = (redcap_data[tcol]<cut).groupby(level='Patient ID').agg(mean)
pat_sum


    

    
Out[8]:




&ltclass 'pandas.core.frame.DataFrame'>
Index: 508 entries, A0001 to A0514
Data columns (total 33 columns):
Admit-Cocaine           507  non-null values
Admit-Cannabinoid       507  non-null values
Test-Cannabinoid        470  non-null values
Admit-Heroin            507  non-null values
Race-Indian             507  non-null values
Admit-Narcotics         507  non-null values
Race-Black              507  non-null values
Vpr                     57  non-null values
Test-Opiates            470  non-null values
V3                      60  non-null values
Admit-None              507  non-null values
Admit-Ritalin           507  non-null values
Admit-Benzodiazapine    507  non-null values
LTR                     454  non-null values
Test-Amphetamines       470  non-null values
Admit-Ecstasy           507  non-null values
Test-Phencyclidine      470  non-null values
Race-Multiple           507  non-null values
Race-Unknown            507  non-null values
Race-Hawaiian           507  non-null values
Admit-Amphetamines      507  non-null values
Race-White              507  non-null values
Test-Barbiturates       470  non-null values
Race-Asian              507  non-null values
VisitNum                508  non-null values
Admit-Other             507  non-null values
Admit-PCP               507  non-null values
Test-Cocaine            470  non-null values
Test-Benzodiazepine     470  non-null values
Tat-1                   57  non-null values
Tat-2                   58  non-null values
%LowVL                  508  non-null values
%LowCD4                 508  non-null values
dtypes: float64(32), int64(1)

In [11]:

    

twpats = ['A0014','A0037','A0041','A0052','A0071','A0110','A0128','A0278','A0281','A0365']
pat_sum.ix[twpats][['LTR', 'V3']]


    

    
Out[11]:






  

    

      

      
LTR
      
V3
    
  
  

    

      
A0014
      
 1
      
NaN
    
    

      
A0037
      
 1
      
  1
    
    

      
A0041
      
 1
      
  1
    
    

      
A0052
      
 1
      
NaN
    
    

      
A0071
      
 1
      
NaN
    
    

      
A0110
      
 1
      
  1
    
    

      
A0128
      
 1
      
  1
    
    

      
A0278
      
 1
      
NaN
    
    

      
A0281
      
 1
      
NaN
    
    

      
A0365
      
 1
      
NaN
    
  

In [13]:

    

non_users = pat_sum[(pat_sum[test_drug_cols]==0).all(axis=1)]

PN_pats = non_users.drop(test_drug_cols, axis=1)


    

In [15]:

    

#PN_pats.to_excel('/home/will/HIVTropism/PN_pats.xlsx')


    

In [20]:

    

import pickle

with open('/home/will/HIVTropism/trop_dict.pkl') as handle:
    trop_data = pickle.load(handle)
tmp = list(trop_data.items())


    

In [17]:

    

for key, val in tmp:
    if key.startswith('A0'):
        print key, val


    

    




---------------------------------------------------------------------------
AttributeError                            Traceback (most recent call last)
<ipython-input-17-3885af77b414> in <module>()
      1 for key, val in trop_data.items():
----> 2     if key.startswith('A0'):
      3         print key, val

AttributeError: 'numpy.int64' object has no attribute 'startswith'

In [27]:

    

(PN_pats['VisitNum']>=3).sum()


    

    
Out[27]:





105

In [25]:

    

len(PN_pats)


    

    
Out[25]:





275

In [ ]: