compositioncalc2.py from Practical Computing for Biologists by Steven H. D. Haddock and Casey W. Dunn. Posted as a Gist by Wayne Decatur (fomightez) with full credit and reference to the original authors and specifying where they freely share the code online. You can see a static IPython Notebook version at http://nbviewer.ipython.org/6102154

compositioncalc2.py

# code by Steven H. D. Haddock and Casey W. Dunn as described in:
# Practical Computing for Biologists
# Steven H. D. Haddock and Casey W. Dunn
# Published in 2011 by Sinauer Associates.
# ISBN 978-0-87893-391-4
# http://www.sinauer.com/practical-computing-for-biologists.html
# see practicalcomputing.org
#
#scripts freely available by the original authors at practicalcomputing.org
#DIRECT LINK: http://practicalcomputing.org/files/pcfb_examples.zip
#
#! /usr/bin/env python
DNASeq = "ATGTCTCATTCAAAGCA"  
SeqLength = float(len(DNASeq))	

BaseList = list(set(DNASeq))	
for Base in BaseList:	
	Percent = 100 * DNASeq.count(Base) / SeqLength	
	print "%s: %4.1f" % (Base,Percent)
 
 
 
 
 
 
 
 
 
#--------------------------------------------------------#
#****ALL BELOW ADDED BY WAYNE DECATUR TO HELP  *********#
#****WITH UNDERSTANDING THIS SCRIPT AND EXPLORING******#
#****FURTHER USING THE INTERACTIVE CONSOLE.    ********#
#--------------------------------------------------------#
print "\n\n\n\n---------------------------------------------------------------------------"
print "Following added by Wayne Decatur beyond Haddock and Dunn code, as a starting point"
print "for understanding the script in this gist."
print "---------------------------------------------------------------------------"
print "The compositoncalc2 script calculates percent of the bases or amino acids in a "
print "DNA or protein sequence,handling nonstandard bases or amino acids. (Actually  "
print "it will calculate percent of any character in a string of characters.)"
print "This improvement over the compositoncalc1 script is made possible by "
print "defining 'BaseList = list(set(DNASeq))'"
print "\nHere, the DNA sequence (DNASeq) = %s" % DNASeq
for Base in BaseList:
  Percent = 100 * DNASeq.count(Base) / SeqLength
  print "%s: %4.1f%%" % (Base,Percent)
 
 
#-----defining the calculator as a function to make interaction via the interactive gist console easier 
def calc (MySeq):
  SeqLength = float(len(MySeq))
  ComponentList = list(set(MySeq))
  for Component in ComponentList:
    Percent = 100 * MySeq.count(Component) / SeqLength
    print "%s: %4.1f%%" % (Component,Percent)
    
#initialize MySeq with same sequence 'calc(MyDNASeq)' works for testing purposes without necesarily needing to define
MySeq = DNASeq
  
 
import pprint  #this is simply to allow use of the pprint command mentioned below
print "\n\nSee the raw code by clicking the blue numbers after 'Gist ID and link' "
print " in the side panel. Tip: Opening the code in a separate browswer window can "
print "make it easier to follow along. See a static IPython Notebook of the input and "
print "output at http://nbviewer.ipython.org/6102154"
 
print "\n\nTo calculate composition for your own sequence (DNA or protein):"
print "  Define the MySeq variable to what you want by typing 'MySeq=\"YOUR SEQUENCE HERE\" ' where "
print "  you type your sequence in place of YOUR SEQUENCE HERE and hit return. Then type, 'calc (MySeq)'"
print "  and hit return."
print "  You can also skip the MySeq defining and type 'calc(\"YOUR SEQUENCE HERE\") and hit return."
print"\nHelp getting started exploring the code in the iPython interactive shell:"
print "  You can get a list of the variable and values by typing 'pprint.pprint(locals())' at the prompt."
print "  Type the name of a variable or object followed by a '?' to get information."
print "In all cases you actually type the double quotes and simply what is between the single quotes."