NucleoSea of Nucleotides

#!/usr/bin/env python2

"""Computes and prints statistics on a 'sea' of nucleotides (poor pun intended - don't judge me).

Usage:
    python nucleo.py [path]
"""
import json
from pprint import pprint
import sys

class NucleoSea():
    """Computes and prints statistics on a 'sea' of nucleotides.

Nucleotides can be added to a NucleoSea one sequence at a time or an entire file at a time.  All symbols are stored
    in a dictionary whose key is the position of the given symbol.  The first position of a symbol in a nucleotide is
    one (not zero).

Attributes:
        pos_map: mappings of 1-indexed positions to symbols in that position
        dec_precision: decimal place precision of output
    """

def __init__(self):
        """Initializes instance."""
        self.pos_map = {}
        self.dec_precision = 2

def add_nucleotide_file(self, path):
        """Reads nucleotide sequences from a file into memory.

Args:
            path: Name of file containing nucleotide sequences.
        Raises:
            IOError: An error occurred while trying to read the file.
        """
        try:
            with open(path, 'r') as f:
                for line in f:
                    self.add_tide(line.rstrip('\n'))
        except:
            print 'Could not read file: ' + path

def add_tide(self, seq):
        """Stores the symbols of a single nucleotide sequence in a position/symbol map.

Args:
            seq: Nucleotide sequence.
        """
        for i in range(len(seq)):
            self.pos_map.setdefault(i + 1, []).append(seq[i])

def print_occurrences(self):
        """Prints all occurrences of symbols at each position (for testing)."""
        pprint(self.pos_map)

def print_percentages(self):
        """Prints position occurrence ratios for all combinations of positions and symbols (in JSON format)."""
        output_map = {}
        for pos, symbols in self.pos_map.items():
            pct_map = {}
            num_sym = float(len(symbols))
            for s in set(symbols):
                pct_map[s] = round(float(symbols.count(s) / num_sym), self.dec_precision)
            output_map[pos] = pct_map
        print json.dumps(output_map, sort_keys=True, indent=4, separators=(',', ' : '))

def main():
    """Adds nucleotides to a NucleoSea instance and prints position occurrence ratios."""
    sea = NucleoSea()
    sea.add_tide('GATAGGAGTAGTGAGT')
    sea.add_tide('GTAGTAGAGTATGATAGTGTA')
    sea.add_tide('GATTAGATGATGATG')
    sea.add_tide('GATATATAGATATATTAGAT')
    sea.add_tide('GATAGATAGT')
    sea.add_tide('GATTAAGATATGATAGTAG')
    sea.add_tide('GATTAGATAGTAGTAGT')
    sea.add_tide('GTATAGATAGTAGTAGTGATGA')
    sea.add_tide('GTAGATGATGATAGTAGTAGT')
    sea.add_tide('GAAGTAGTGATGAGTAG')
    #sea.print_occurrences()
    sea.print_percentages()

if __name__ == '__main__':
    main()

Python Fiddle

Python Cloud IDE