#get-mean-log2.py
#This program takes 4 command line options:
#A file of log2 value, chromosome name, and coordinate begin and end
#it prints out the number of probes in the
#indicated interval and their  average and mean log2 values

import sys
import math

#return mean value of entries in a list
def calcMean(values):
  n=0
  sum=0
  for i in values:
    sum+=i
    n+=1
  return float(sum)/float(n)
  
#return standard deviation of entries in a list
def calcStdDev(values):
  mean=calcMean(values)
  sumMeanDiffSqrd=0
  for i in values:
    sumMeanDiffSqrd+=(i-mean)**2
  stdDev=math.sqrt(sumMeanDiffSqrd/len(values))
  return stdDev

#read command line options
fileName=sys.argv[1]
chrName=sys.argv[2]
chrStart=int(sys.argv[3])
chrEnd=int(sys.argv[4])

myFile=open(fileName,'r')
values=[]
n=0
for line in myFile:
  line.rstrip()
  (chr,probePos,log2)=line.split()
  probePos=int(probePos)
  log2=float(log2)
  #only consider probes on the right chromosome
  #and within the indicated interval
  if chr == chrName and probePos >= chrStart and probePos <= chrEnd :
    values.append(log2)
    n+=1
myFile.close

numProbes=len(values)
mean=calcMean(values)
stdDev=calcStdDev(values)

print numProbes,'probes in interval'
print 'mean: ',mean
print 'standard deviation',stdDev