Integration of data using Simpson's rule

#! /usr/bin/env python
import glob
import os
import fileinput
print "<SimpInt.py> a script to quickly calculate the area for a set"
print "of analytes. Written by Bas Jansen, Leiden University Medical"
print "Center, March 20th 2013."
print ""

# Simpson's rule
def integrate(y_vals,h):

The recommendation for python is four spaces for indentation, you appear to be using 2

  i=1
  total=y_vals[0]+y_vals[-1]

It's also recommended to have spaces around binary operators

  for y in y_vals[1:-1]:

Use for i, y in enumerate(y_vals, 1): rather then counting your own i value

    if i%2 == 0:
      total+=2*y
    else:
      total+=4*y
    i+=1
  return total*(h/3.0)

# Reads the reference file
ref=[]

I'd recommended not using an abbreviation her

for refs in glob.glob("*.ref"):
  fr=open(refs)

It's recommended to use a with block:

  with open(refs) as fr:
      do stuff with file
  file will be automatically closed here

That ensures your file will be closed even if an exception occours

  for line in fr.readlines():
    parts=line.rstrip('\n').split('\t')
    ref.append(parts)

You shouldn't really store the header ref here.

  fr.close()



# Initialize the output file
f=open('output.xls','w')
for i in range(1,len(ref)):
  f.write(ref[i][0]+"\n")
f.close()



# Find XY files in current directory
xy_files=[]
for files in glob.glob("*.xy"):
  xy_files.append(files)
  x=[]
  area=[]
  print "Parsing: "+files
  fh=open(files)
  for line in fh.readlines():

Actually this is the same as for line in fh:

    y=[float(i) for i in line.split()]
    x.append(y)
  fh.close()

  # Pick the XY window
  print "Progress indicator:"
  print "0%"
  for i in range(1,len(ref)):

Whenever possible, don't iterate over indexes. Instead iterate over the lists. So something like for line in ref[1:]: If you really need the indexes us enumerate

    obs=[]
    vals=0
    low=float(ref[i][1])-float(ref[i][3])
    high=float(ref[i][1])+float(ref[i][3])
    # Progress indicator
    if i %(len(ref)/10) == 0:
      print str(i/(len(ref)/10))+"0%"

Why divide by 10 just to add the 0 back?

    for j in range(0,len(x)):

Use for j in x: and then avoid the indexing

      if x[j][0] > low and x[j][0] < high:
        coords=[float(l) for l in x[j]]
        obs.append(coords)
        vals+=1

What's this for?

    if obs:
      y_vals=[]
      h=(obs[len(obs)-1][0]-obs[0][0])/len(obs)
      for k in range(0,len(obs)):
        y_vals.append(obs[k][1])
      area.append(integrate(y_vals,h))

  # Neat trick to append the area from other input files
  l = 0
  for line in fileinput.input('output.xls',inplace=1):

If you need to iterate over multiple sources at once, use a zip.

    print "%s\t%s\n" % (line.rstrip(),area[l]),
    l+=1

Your neat trick is what's killing you. This is designed for processing the output of some other program. Its not good for continually adding to the end of lines you have already written. What you should do is store everything in a big list and then output it at the end.

# Display the legend (& filenames) at bottom of output file
f=open('output.xls','a')
legend = "Peak"
for m in range(0,len(xy_files)):
  legend+="\t"+str(xy_files[m])
legend+="\n"
f.write(legend)
f.close()

# Being polite
raw_input("Finished, press any key to exit")