def run_paml(infile, treefile):
cml = codeml.Codeml(alignment = infile, tree = treefile, \
out_file = "results.out", working_dir = "./temp")
cml.read_ctl_file("alt_codeml.ctl")
#cml.print_options()
fore = cml.run()
fore_dat = ((fore.get('NSsites')).get(2))
fore_lnL = fore_dat.get('lnL')
fore_omegas = (fore_dat.get('parameters')).get('site classes')
fore_omega = ((fore_omegas.get(2)).get('branch types')).get('foreground')
cml.read_ctl_file("null_codeml.ctl")
#cml.print_options()
null = cml.run()
null_lnL = ((null.get('NSsites')).get(2)).get('lnL')
df = 1
chisq = 2 * (fore_lnL - null_lnL)
try:
pval = cdf_chi2(df, chisq)
except:
pval = 1.000
return fore_omega, chisq, pval
def testCdfChi2(self):
self.assertRaises(ValueError, chi2.cdf_chi2, df = 0, stat = 3.84)
self.assertRaises(ValueError, chi2.cdf_chi2, df = 1, stat = -3.84)
self.assertRaises(TypeError, chi2.cdf_chi2, df = "d", stat ="stat")
self.assertAlmostEqual(chi2.cdf_chi2(2, 3.84), 0.1466070, places=5)
print "\nThe sum of all posterior probabilities should equal unity. The actual sum is:", (sum(posterior)) if sum(posterior) == 1: print "\nSummation checks out!\n" else: print "Sum does not equal 1. Something went wrong...or maybe not. Remember that these are high-precision libraries, and 0.99999... is acceptable.\n" print "Traditional likelihood ratio tests:\n" #teststat = -2 * log ( likelihood1 / likelihood 0) print "M1a vs. M2a\n" df = 2 print "The two likelihoods are:", round(floatlike[1], 2), round(floatlike[2], 2) teststat = abs(2 * (-1 * round(floatlike[2], 2) - -1 * round(floatlike[1], 2))) print "The test statistic is:", teststat m12_p = cdf_chi2(df, float(teststat)) print "p-value:", cdf_chi2(df, float(teststat)) print "\nM7 vs. M8\n" df = 2 print "The two likelihoods are:", round(floatlike[7], 2), round(floatlike[8], 2) teststat = abs(2 * (-1 * round(floatlike[8], 2) - -1 * round(floatlike[7], 2))) print "The test statistic is:", teststat m78_p = cdf_chi2(df, float(teststat)) print "p-value:", cdf_chi2(df, float(teststat)) if args.M8a_test: print "\nM8a vs. M8\n" df = 1
#!/usr/bin/env python
import sys
from Bio.Phylo.PAML.chi2 import cdf_chi2
# This script calculates the p-value for the likelihood ratio test give two
# likelihoods and degrees of freedom.
def usage():
print "LRT.py <Likelihood 1> <Likelihood 2> <degrees of freedom>"
if len(sys.argv) != 4:
usage()
sys.exit()
lnL1 = float(sys.argv[1])
lnL2 = float(sys.argv[2])
degrees = int(sys.argv[3])
LRTstat = 2*(lnL1 - lnL2)
print LRTstat
p = cdf_chi2(degrees, LRTstat)
print p
tree, loglk = wrappers.phyml(a)
return (a, tree)
def run_paml(a, tree, alignName, outfile, nonsignificantFile, neutralFile):
try:
codemlInstance = wrappers.Codeml(a, tree)
neutral = codemlInstance.fit("M1a")
positive = codemlInstance.fit("M2a")
except ValueError, e:
print (alignName, e)
return
fitDict = {}
n = True
LRTstat = 2*(positive["lnL"] - neutral["lnL"])
if LRTstat > 0:
p = cdf_chi2(2, LRTstat)
if p < 0.05:
n = False
if not n:
outfile.write("%s\t%f\n" % (alignName, p))
else:
nonsignificantFile.write("%s\t%f\n" % (alignName, p))
else:
p = cdf_chi2(2, LRTstat*-1)
neutralFile.write("%s\t%f\n" % (alignName, p))
alignment, directory = get_arguments(sys.argv[1:])
alignDict = {}
# Check if alignment or directory was given and calculate stats accordingly
def testCdfChi2(self):
self.assertRaises(ValueError, chi2.cdf_chi2, df=0, stat=3.84)
self.assertRaises(ValueError, chi2.cdf_chi2, df=1, stat=-3.84)
self.assertRaises(TypeError, chi2.cdf_chi2, df="d", stat="stat")
self.assertAlmostEqual(chi2.cdf_chi2(2, 3.84), 0.1466070, places=5)
# This script returns the p-value from chi2 distributions
# for likelihood ratio tests
# Imports
from Bio.Phylo.PAML.chi2 import cdf_chi2
df = 1
statistic = 24.26
chi2 = cdf_chi2(df, statistic)
print ("Chi2:",chi2)
expansion_sfs_file.write(str(expansion_sfs[i]) + '\n')
expansion_sfs_file.close()
# Output SFS for growth model
growth_sfs = dadi.Inference.optimally_scaled_sfs(growth_model, data)
growth_sfs_file = open("growthModelSFS.txt", 'w')
for i in range(1,len(growth_sfs)-1):
growth_sfs_file.write(str(growth_sfs[i]) + '\n')
growth_sfs_file.close()
if expansion_ll_opt > growth_ll_opt:
print "Testing significance of expansion..."
LRTstat_neutral = 2*(expansion_ll_opt - neutral_ll)
degrees = len(expansion_params)
print "LRT Statistic- Neutral:", LRTstat_neutral
p_neutral = cdf_chi2(degrees, LRTstat_neutral)
print "p-value neutral =",p_neutral
LRTstat_growth = 2*(expansion_ll_opt - growth_ll_opt)
print "LRT Statistic- Exponential Growth:", LRTstat_growth
p_growth = cdf_chi2(degrees, LRTstat_growth)
print "p-value growth =",p_growth
if p_neutral < 0.05:
print "Working on likelihood surface..."
likelihood_grid(expansion_func_ex, data, ns, pts_l, "expansion")
if p_growth > 0.05:
likelihood_grid(growth_func_ex, data, ns, pts_l, "growth")
else:
print "Testing significance of exponential growth..."
LRTstat_neutral = 2*(growth_ll_opt - neutral_ll)
degrees = len(growth_params)
#!/usr/bin/env python
import sys
from Bio.Phylo.PAML.chi2 import cdf_chi2
# This script calculates the p-value for the likelihood ratio test give two
# likelihoods and degrees of freedom.
def usage():
print "LRT.py <Likelihood 1> <Likelihood 2> <degrees of freedom>"
if len(sys.argv) != 4:
usage()
sys.exit()
lnL1 = float(sys.argv[1])
lnL2 = float(sys.argv[2])
degrees = int(sys.argv[3])
LRTstat = 2 * (lnL1 - lnL2)
print LRTstat
p = cdf_chi2(degrees, LRTstat)
print p
# 1.4 Results results = .read() dict.get(key) #returns a specific dictionary result, based on desired key # 2.0 CodeML cml = codeml.Codeml() cml.alignment = "align.phylip" #tells codeML the alignment name cml.tree = "species.tree" #tells codeML the tree name cml.out_file = "results.out" #sets output file name cml.working_dir = "./scratch" #sets working directory to store output files cml.run() # 3.0 BaseML bml = baseml.Baseml() # >Note: use same methods/attributes as CodeML, just replace with 'bml' # 4.0 yn00 yn = yn00.Yn00() # >Note: use same methods/attributes as CodeML, replace with 'yn' # >Note: you do not need a tree file # 5.0 Chi-Squared df = 2 #sets degrees of freedom statistic = 7.21 #sets chi2 test statistic cdf_chi2(df, statistic) #returns chi2 value
