plotDist

#!/usr/bin/python

import sys

if len(sys.argv) < 3:
    print 'USAGE: plotDist FILENAME INFILE [[INFILE] ...]'
    sys.exit(1)

fname = sys.argv[1]
infiles = sys.argv[2:]

logmethods = ['rsat', 'nhmmer', 'mast']

import math

meth = {}

import numpy as np

# Quick read to know which methods are there
for f in infiles:
    for l in open(f):
        s = l.rstrip().split('\t')
        me = s[9]
        meth[me] = meth.get(me, None)

def iterScores(method, infiles):
    for f in infiles:
        for l in open(f):
            s = l.rstrip().split('\t')
            sc = float(s[10])
            me = s[9]
            
            if me == method:
                if me in logmethods:
                    yield -math.log10(sc)
                else:
                    yield sc

for me in meth:
    meth[me] = np.fromiter(iterScores(me, infiles), np.float)
    
import matplotlib
matplotlib.use('Agg')
import matplotlib.pyplot as plt
    
fig = plt.figure(figsize=(10,10))

colors = ['b', 'r', 'g', 'k']

import numpy as np
nmeth = np.array(meth.keys())

# Array reshape (tricky!)
# 1- Finger crossed for a perfect square
square = np.sqrt( len(nmeth) )
if square.is_integer():
    # Yep
    rows = cols = int(square)
else:
    # Oh snap!
    cols = int(square)
    rows = len(nmeth) / float(cols)
    if not rows.is_integer():
        # Some fake signals will be added
        rows = int(rows) + 1
    rows = int(rows)

i = 0
for m in sorted(meth.keys()):
    ax = fig.add_subplot(rows, cols, i)
    
    j = i
    while True:
        if j > len(colors)-1:
            j = i-len(colors)
        else:
            c = colors[j]
            break
    
    ax.hist(meth[m], bins=100, color=c)
    
    ax.set_title('%s'%m)
    #ax.set_ylim(0, 1)
    #ax.set_xlim(min(pdist),0)
    
    i += 1

plt.tight_layout()
plt.savefig('%s.png'%fname)
plt.savefig('%s.pdf'%fname)