comparison hetbox.py @ 4:1c21e9f3104b draft default tip

updated script comments and remove "evaluation" column of the report

3:479f860eb8a9

#!/usr/bin/env python
# Code by Boris Rebolledo-Jaramillo
# (boris-at-bx.psu.edu)
# Edited by Nick Stoler
# (nick-at-bx.psu.edu)
# New in this version:
# - Add in proper header line if not present

import os
import sys
import array
import numpy

           'outpch':"*",'main':"Distribution of minor allele frequencies",
           'cex.lab':"1.5"}
p = graphics.boxplot(formula, axes=0,ylim=ylimit, lty=1,**kwargs1)

table = base.table(DATA.rx2('SAMPLE'))
#graphics.text(-1, 1, 'N:', font=2)
for i in range(1, base.length(table)[0]+1, 1):
    graphics.text(i, -1, table[i-1], font=2)

graphlabels = base.names(table)
kwargs3 = {'pos':"-2", 'las':"2", 'cex.axis':"1"}
graphics.axis(1, at=range(1, len(graphlabels)+1, 1),labels=graphlabels, **kwargs3)
graphics.axis(2,at=(range(0,60,10)),pos=0,font=2)
grdevices.dev_off()

if not report:
  sys.exit(0)


####################################
# GENERATE REPORT
# report should be something like:
# SAMPLE NoHET MEDIAN MAD TEST 
# s1     7      10%   n   p/w/f
# n <= 5      pass
# 6 <= n <=10 warn
# n >= 11     fail
# MAD <= 2.0  fail
# MAD > 2.0    pass
###################################

SAMPLES=[]
for i in range(len(table)):
    SAMPLES.append(base.names(table)[i])

    MEDIAN = numpy.median(VALUES)
    MAD    = numpy.median([abs(i - MEDIAN) for i in VALUES]) # Median absolute distance (robust spread statistic)
    return [NoHET,MEDIAN, MAD]

boxreport = open(report, "w+")
boxreport.write("SAMPLE\tTOTAL.SITES\tMEDIAN.FREQ.\tMAD.FREQ\tEVAL\n")
for sample in SAMPLES:
    ENTRY = [sample] + boxstats(infile,sample)
    if ENTRY[1] <= 5:
        ENTRY.append('pass')
    elif 6 <= ENTRY[1] <=10:
        ENTRY.append('warn')
    elif ENTRY[1] >= 11:
        ENTRY.append('fail')
    if ENTRY[3] <=2.0:
        ENTRY.append('fail')
    elif ENTRY[3] >2.0:
        ENTRY.append('pass')
    if len(set(ENTRY[4:6])) == 2:
        ENTRY.append('warn')
    else:
        ENTRY.append(list(set(ENTRY[4:6]))[0])
    boxreport.write ('%s\t%d\t%.1f\t%.1f\t%s\n' % tuple([ENTRY[i] for i in [0,1,2,3,6]]))

boxreport.close()

4:1c21e9f3104b

#!/usr/bin/env python

# Code by Boris Rebolledo-Jaramillo
# (boris-at-bx.psu.edu)
# Edited by Nick Stoler
# (nick-at-bx.psu.edu)
# New in this version:
# - Add in proper header line if not present


import os
import sys
import array
import numpy

           'outpch':"*",'main':"Distribution of minor allele frequencies",
           'cex.lab':"1.5"}
p = graphics.boxplot(formula, axes=0,ylim=ylimit, lty=1,**kwargs1)

table = base.table(DATA.rx2('SAMPLE'))
graphics.text(0, -1, 'N:', font=2)
for i in range(1, base.length(table)[0]+1, 1):
    graphics.text(i, -1, table[i-1], font=2)

graphlabels = base.names(table)
kwargs3 = {'pos':"-2", 'las':"2", 'cex.axis':"1"}
graphics.axis(1, at=range(1, len(graphlabels)+1, 1),labels=graphlabels, **kwargs3)
graphics.axis(2,at=(range(0,60,10)),pos=0,font=2)
grdevices.dev_off()

if not report:
    sys.exit(0)


SAMPLES=[]
for i in range(len(table)):
    SAMPLES.append(base.names(table)[i])

    MEDIAN = numpy.median(VALUES)
    MAD    = numpy.median([abs(i - MEDIAN) for i in VALUES]) # Median absolute distance (robust spread statistic)
    return [NoHET,MEDIAN, MAD]

boxreport = open(report, "w+")
boxreport.write("#sample\tNo.sites\tmedian.freq\tMAD.freq\n")

for sample in SAMPLES:
    ENTRY = [sample] + boxstats(infile,sample)
    boxreport.write ('%s\t%d\t%.1f\t%.1f\n' % tuple([ENTRY[i] for i in [0,1,2,3]]))
boxreport.close()