Baseanno.py

############################################################################################################################# #baseanno.py [options] <text file with annotations and start/end indices> # #Example: python baseanno.py -p myAnnotations.txt # #Version: 1.0 Last Modified: 12/22/2009 Author: Zachary S. L. Foster # #---------------------------------------------------------------------------------------------------------------------------# #Argument: a whitespace-delimitated text (.txt) file which contains a list of annotations in the following format: # # Annotation Start index End index ... # # *The name of the annotation have no spaces; if it does, use the -t modifier # #---------------------------------------------------------------------------------------------------------------------------# #>Outputs a tab-delimitated list of base indices and their respective annotations # #>If a single base has more than one annotation, then the additional annotations are added after the first on the same line # #---------------------------------------------------------------------------------------------------------------------------# #Modifiers: # # -d  : Save debug log to current working directory # # -p  : Print output to standard out (usually the shell/screen being called from) # # -t  : Use if the name of your annotations have spaces AND the input data is tab-delimitated # ############################################################################################################################# ###Imports / Variable Initilization########################################################################################## import os, string, sys defArgList = ['baseanno.py','-p','-t','C:/Python26/baseannoTestData.txt'] #argument list used during script testing the with Python GUI IDLE argList = sys.argv #argument list supplied by user argNum = len(argList) #the number of arguments supplied minArgNum = 1 #the smallest amount of arguments with which it is possible to run the script saveDubug = False #is True if the debug is to be saved warning = False #is set to 'True' if the program encounters any minor errors during the analysis; recorded in debug log tabDelim = False #is 'True' if the -t modifier is supplied; enforces tab-delimitation in input file helpOnly = False #is 'True' when no arguments are given; only help menu is printed debugLog = ['*********DEBUG LOG*********\n'] #where all errors/anomalies are recorded; saved if the -d modifier is supplied savePath = './' #the directory in which the output will be saved defSavePath = 'C:/Python26/' #save path used during script testing with the python GUI IDLE inData = [] #a list of lists containing the data from the input file [[ref,density,actual],...] parsedData = [] #a list containing data for each 1000 bp chunk [[GSS ref match length...see above]...] outData = [] #a list of strings corresponding to each line in the output file modifiers = [] #eventually contains a list of all modifiers and their arguments supplied by the user allMods = 'dpt' #all the modifiers recognized by the script activeMods = '' #a string containing all modifiers specified by the user modArgs = [] #arguments supplied for the modifiers; assumed in the same order as the modifiers ############################################################################################################################# def printHelp(): print '/---------------------------------------------------------------------------------------\\' print '| baseanno.py [options] <text file with annotations and start/end indices> |' print '|---------------------------------------------------------------------------------------|' print '| Example: python baseanno.py -p myAnnotations.txt |' print '|---------------------------------------------------------------------------------------|' print '| Version: 1.0 Last Edited: 12/22/2009 |' print '| >Outputs a tab-delimitated list of base indices and their respective annotations |' print '| >If a single base has more than one annotation, then the additional annotations are |' print '| added after the first on the same line |' print '|---------------------------------------------------------------------------------------|' print '| Modifiers: |' print '| -d  : Save debug log |' print '| -p  : Print output to standard out (usually the shell being called from) |' print '| -t  : Use if your annotations have spaces AND the input data is tab-delimitated |' print '\\---------------------------------------------------------------------------------------/' #Error handling function##################################################################################################### #>Is called when the script encounters a fatal error # #>Prints the debug log to standard out (usually the screen) # def errorExit(): print 'The program was forced to exit prematurely, printing debug log...\n' for line in debugLog: print line sys.exit() ############################################################################################################################# ###Argument Interprtation#################################################################################################### #>Parses the arguments supplied by the user, or the default values if the script is being run on IDLE # #>If no arguments are given, the help menu is printed # #>Modifiers and their arguments are isolated from the raw input for later processing (in Modifier Interpretation) # if __name__ == '__main__': #If the program is being called independent of the Python GUI, IDLE... if argNum > minArgNum: #If at least the minimum number of arguments necessary is supplied... if os.path.exists(argList[-1]) == 0: #if the path dose not exist debugLog.append('Error: Invalid file path to input data') errorExit() #end the program elif argNum == 1: #If no arguments are supplied... helpOnly = True printHelp() #prints help menu else: debugLog.append('Error: Too few arguments supplied\n') errorExit() else: #If the script is being imported on to IDLE argList = defArgList #use default arguments argNum = len(defArgList) if argNum == 1: #If no arguments are supplied... helpOnly = True printHelp() #prints help menu savePath = defSavePath #sets the save path to the default, specified in the variable initialization section debugLog.append('Alert: default debugging input arguments are being used\n') if helpOnly == False: #if arguments were supplied... inPath = argList[-1] #the path to the fasta file containing the input data if argNum > minArgNum + 1: #if modifiers are present (i.e. more the minimum number of arguments) modifiers = argList[1:-minArgNum] #everything before the required arguments are modifiers and their arguments ############################################################################################################################# if helpOnly == False: ###Modifier Interpretation############################################################################################### #>Parses any modifiers and modifier arguments determined by the previous section of code, "Argument Interpretation" # #>Given arguments are compared against a list of known arguments # #>Matches found change the appropriate variable for the desired effect of the modifier the script # if len(modifiers) > 0: #if modifiers are supplied for mod in modifiers: #loops through the list of modifiers and modifier arguments if mod[0] == '-' and len(mod) == 2: #list entry considered modifier if it starts with - and is only two characters activeMods += mod[1:] #sorts the modifiers into activeMods... else: modArgs.append(mod) #assumes everything else to be a modifier argument for letter in activeMods: #loops through supplied modifiers if string.find(allMods,letter) == -1: #checks if the modifier is recognized by this script debugLog.append('Warning: Unexpected modifier: ' + letter + '\n') warning = True #if the input modifier is not found else: #if it is a recognized modifier... if letter == 'd': #if -d is supplied... saveDebug = True #The debug log will be saved to the current working directory if letter == 'p': #if -p is supplied... printOut = True #The results will be printed to the standard output (usually the screen) if letter == 't': #if -t is supplied... tabDelim = True #tab-delimitation is enforced for input file ######################################################################################################################### ###Input data parseing and file procedures############################################################################### inHandle = open(inPath, 'r') #creates a file object for line in inHandle: #loops through the lines of the input file if tabDelim: lineParts = string.split(line,'\t') #produces a list of all the contents of each line (delimitated by tabs only) else: lineParts = string.split(line) #produces a list of all the contents of each line (delimitated by spaces/tabs) if len(lineParts) >= 3: annotation = lineParts[0] start = int(lineParts[1]) end = int(lineParts[2]) inData.append([annotation,start,end]) #adds parsed line to the list that will be analyzed inHandle.close() #closes the file object ######################################################################################################################### ###Consensous creation################################################################################################### outData = [] lastBaseIndex = 0 for anno in inData: if anno[2] > lastBaseIndex: lastBaseIndex = anno[2] for index in range(1,lastBaseIndex + 1): outData.append(str(index)) for anno in inData: start = anno[1] end = anno[2] for index in range(start - 1,end): outData [index] += '\t' + anno[0] ######################################################################################################################### ###Out file writing and saveing procedures############################################################################### savePath += os.path.basename(inPath)[:-4] + '_byBase.txt' #the path to where the output is saved outHandle = open(savePath, 'w') #opens the file object for saving the output for line in outData: #prints every line of outData to the output file... outHandle.write(line + '\n') outHandle.close() if printOut: #if the -p modifier is supplied for line in outData: #prints every line to the standard output... print line ######################################################################################################################### ###Debug Saving procedures############################################################################################### if saveDebug: #if the -d modifier is supplied... DebugSavePath = savePath[:-4] + '.BADebug.txt' #the path to where the debug log is to be saved debugHandle = open(fileSavePath, 'w') #opens the file object for saving the debug log for line in debugLog: debugHandle.write(line) debugHandle.close() #########################################################################################################################