CODEWITHSUNDEEP
python
clintval
clintval

Reputation: 369

How to bin many category combinations (72 variables) in python lists

I have data stored in a list of lists organized like so:

lst = [
      ['FHControl', G, A]
      ['MNHDosed', G, C]
      ]

For row in lst: row[0] there are a total of 12 categories (I've listed two in the sample code above). For row[1] and row[2] I am only concerned with 6 of the combinations of these letters. Therefore, I have 72 possible combinations of this data per row in lst and need to count the instances of each combination without having to write dozens of nested if loops.

I am attempting in creating two functions to parse through these lists and bin the incidences of these 72 combinations. How can I use two function like what I am beginning to write below to update these variables? Do I need to construct the dictionaries as class variables so that I can continue to update them as I iterate through both functions? Any guidance would be great!

Here is the code I have currently that initializes all 72 variables into 6 dictionaries (for the 6 combinations of letters in row[1] and row[2]):

def baseparser(lst):
    TEMP = dict.fromkeys('FHDosed FHControl FNHDosed FNHControl '
                         'FTDosed FTControl MHDosed MHControl '
                         'MNHDosed MNHControl MTDosed MTControl'.split(), 0)
    TRI_1, TRI_2, TRV_1, TRV_2, TRV_3, TRV_4 = ([dict(TEMP) for i in range(6)])

    for row in lst:
        if row[0] == 'FHDosed':
            binner(row[0], row[1], row[2])
        if row[0] == 'FHControl':
            binner(row[0], row[1], row[2])
        etc.

def binner(key, q, s):
    if (q == 'G' and s == 'A') or (q =='C' and s =='T'):
        TRI_1[key] += 1
    elif (q == 'A' and s == 'G') or (q =='T' and s =='C'):
        TRI_2[key] += 1
    elif (q == 'G' and s == 'T') or (q =='C' and s =='A'):
        TRV_1[key] += 1
    elif (q == 'G' and s == 'C') or (q =='C' and s =='G'):
        TRV_1[key] += 1
    elif (q == 'A' and s == 'T') or (q =='T' and s =='A'):
        TRV_1[key] += 1
    elif (q == 'A' and s == 'C') or (q =='T' and s =='G'):
        TRV_1[key] += 1

Upvotes: 2

Views: 268

Answers (1)

unutbu
unutbu

Reputation: 880547

Your code could be simplified to:

TEMP = dict.fromkeys('''FHDosed FHControl FNHDosed FNHControl FTDosed FTControl MHDosed 
                      MHControl MNHDosed MNHControl MTDosed MTControl'''.split(), 0)
TRI_1, TRI_2, TRV_1, TRV_2, TRV_3, TRV_4 = [TEMP.copy() for i in range(6)]

dmap = {
    ('G', 'A'): TRI_1,
    ('C', 'T'): TRI_1,
    ('A', 'G'): TRI_2,
    ('T', 'C'): TRI_2,        
    ('G', 'C'): TRV_1,
    ('C', 'G'): TRV_1,        
    ('A', 'T'): TRV_1,
    ('T', 'A'): TRV_1,        
    }

for row in lst:
    key, q, s = row
    dmap[q, s][key] += 1

Another possiblity is to use one dict of dicts instead of 6 dicts:

TEMP = dict.fromkeys('''FHDosed FHControl FNHDosed FNHControl FTDosed FTControl MHDosed 
                      MHControl MNHDosed MNHControl MTDosed MTControl'''.split(), 0)
TR = {key:TEMP.copy() for key in ('TRI_1', 'TRI_2', 'TRV_1', 'TRV_2', 'TRV_3', 'TRV_4')}


dmap = {
    ('G', 'A'): 'TRI_1',
    ('C', 'T'): 'TRI_1',
    ('A', 'G'): 'TRI_2',
    ('T', 'C'): 'TRI_2', 
    ('G', 'C'): 'TRV_1',
    ('C', 'G'): 'TRV_1', 
    ('A', 'T'): 'TRV_1',
    ('T', 'A'): 'TRV_1',        
    }

lst = [
      ['FHControl', 'G', 'A'],
      ['MNHDosed', 'G', 'C']
      ]

for row in lst:
    key, q, s = row
    TR[dmap[q, s]][key] += 1

print(TR)

The advantage of doing it this way is that you have fewer dicts in your namespace, and it may be easier to refactor the code later using a dict of dicts instead of hard-coding 6 dicts.

Following up on Midnighter's suggestion, if you have pandas, you could replace the dict of dicts with a DataFrame. Then the frequency of pairs could be computed using pd.crosstabs like this:

import pandas as pd

dmap = {
    'GA': 'TRI_1',
    'CT': 'TRI_1',
    'AG': 'TRI_2',
    'TC': 'TRI_2', 
    'GC': 'TRV_1',
    'CG': 'TRV_1', 
    'AT': 'TRV_1',
    'TA': 'TRV_1',        
    }

lst = [
      ['FHControl', 'G', 'A'],
      ['MNHDosed', 'G', 'C']
      ]

df = pd.DataFrame(lst, columns=['key', 'q', 's'])
df['tr'] = (df['q']+df['s']).map(dmap)

print(df)
#          key  q  s     tr
# 0  FHControl  G  A  TRI_1
# 1   MNHDosed  G  C  TRV_1

print(pd.crosstab(rows=[df['key']], cols=[df['tr']]))

yields

tr         TRI_1  TRV_1
key                    
FHControl      1      0
MNHDosed       0      1

Upvotes: 4

Related Questions